Icaal 4-1

Pacific Linguistics E-8
Pacific Linguistics is a publisher specialising in grammars and linguistic descriptions, dictionaries and other
materials on languages of the Pacific, Taiwan, the Philippines, Indonesia, East Timor, southeast and south
Asia, and Australia. Pacific Linguistics, established in 1963 through an initial grant from the Hunter Douglas
Fund, is associated with the Research School of Pacific and Asian Studies at The Australian National
University. The authors and editors of Pacific Linguistics publications are drawn from a wide range of
institutions around the world. Publications are refereed by scholars with relevant expertise, who are usually
not members of the editorial board.

FOUNDING EDITOR: Stephen A. Wurm
EDITORIAL BOARD: Malcolm Ross and I Wayan Arka (Managing Editors),
Mark Donohue, Nicholas Evans, David Nash, Andrew Pawley,
Paul Sidwell, Jane Simpson, and Darrell Tryon

Copyright in this edition is vested with the authors
First published 2011
National Library of Australia Cataloguing-in-Publication entry:
Author: International Conference on Austroasiatic Linguistics (4th:, 2009 : Salaya, Thailand).
Title: Austroasiatic studies [electronic resource]: papers from ICAAl 4, Mon-Khmer Studies
Journal Special issue no. 2 / edited by Sophana Srichampa and Paul Sidwell.
ISBN: 9780858836419 (Electronic document)
Subjects: Austroasiatic languages--Congresses.
Other Authors/ Srichampa, Sophana.
Contributors: Sidwell, Paul J.
SIL International.
Australian National University. Research School of Pacific and Asian Studies.
Pacific Linguistics
Mahwitthaylai Mahidon.

Typeset by the editors and assistants

Published concurrently by:

Pacific Linguistics
School of Culture, History and Language
College of Asia and the Pacific
The Australian National University

Research Institute for Languages and Cultures of Asia
Mahidol University
Nakhon Pathom, Thailand

SIL International
7500 West Camp Wisdom Road
Dallas, Texas, USA

iii
Contents

Preface............................................................................................................................................... iv
The Syntax of Khasi Adverbial Clauses............................................................................................................ 1
George BEDELL

Deponent Verbs and Middle-Voice Nouns in Temiar ..................................................................................... 11
Geoffrey BENJAMIN

Toward Proto Pearic: Problems and Historical Implications.......................................................................... 38
Michel FERLUS

Goal-marking in Munda with special reference to the Lower Munda languages ........................................... 52
Arun GHOSH

A Synopsis of Mal Phonetics .......................................................................................................................... 69
THERAPHAN L-Thongkum, CHOMMANAD Intajamornrak, KANITHA Putthasatien,
YUPAPORN Huadsiri, and SUPAPORN Phalipat

Mon Converging Towards Thai Models: ........................................................................................................ 81
Patrick McCORMICK

A Study of Language Use and Literacy Practices to Inform Local Language Literature Development
among Khmu in Thailand................................................................................................................................ 98
Michelle M. MILLER & Timothy M. MILLER

Multiple Uses of the Pronoun de in Kmhmu' ............................................................................................... 112
R. Anne OSBORNE

About aj in Contemporary Khmer. ............................................................................................................ 124
Denis PAILLARD

Pearic, a Dying Branch of Austroasiatic Languages and Its Struggle for Survival ....................................... 138
SUWILAI Premsrirat

Aspects of Ho Phonetics and Phonology....................................................................................................... 154
Anna PUCILOWSKI

Notes on Glottal Constriction in Gorum........................................................................................................ 174
Felix RAU

Comitative PP as Head in Externally-Headed Relative Clauses in Khasi ..................................................... 184
Karumuri v. SUBBARAO & Gracious M. TEMSEN

iv
Preface
History of International Conference on Austroasiatic Linguistics (ICAAL)
The field of Austroasiatic linguistics has had a remarkable history, not the least for having no
regular international conference for decades, even though there was sufficient scholarship to support an
international journal, Mon-Khmer Studies. The field had developed dramatically in the 1960s - with special
being paid attention to Indo-China, there was a tremendous flow of new field data. And it was in this
circumstance that the ICAAL was first mooted in 1972.
The suggestion was first raised by Norman Zide, long time Munda languages scholar based at the
University of Chicago, who had been present at earlier related meetings. These were two gatherings held at
the School of Oriental and African Studies (London) in 1961 and 1965, the papers published as: Linguistic
Comparison in South East Asia (1963, H. L. Shorto, ed.), and Indo-Pacific Linguistic Studies (1965, G. B.
Milner and E. J. A. Henderson eds. Lingua vols. 14 and 15). While these meetings had broad agendas, there
was a strong representation by Austroasiatic, due in large part of scholars such as Harry Shorto and Norman
Zide.
Lawrence Reid, then at Hawaii, took on the main effort of organizing a conference dedicated to
Austroasiatic, and was successful in attracting a substantial grant from the US National Science Foundation
to help bring scholars from around the world to Hawaii. Subsequently, the first ICAAL was held was held at
the University of Hawaii in Honolulu during the first week of January in 1973. It was an historic gathering,
not just a tremendous success as a meeting, but a programmatic turning point for the field. The two
conference volumes that came out in 1976, as Austroasiatic Studies (P. N. Jenner, L. C. Thompson, S.
Starosta eds. Oceanic Linguistics special publications No. 13) presented some 50 papers in 1343 pages. The
contributions established the state of the art with treatments of classification, grammatical description,
phonological and morphological history, and language contact, among others. It was the single most
important publication for Austroasiatic linguistics for the next 30 years, until the appearance of Shortos
posthumous A Comparative Mon-Khmer Dictionary in 2006.
That first meeting also decided to designate Mon-Khmer Studies (MKS) the principle journal of the
field. Previously MKS had been produced intermittently out of Saigon, where it had mainly served SIL
associated scholars. Subsequently the journal moved to Hawaii and a broader editorial board established.
Today MKS is based at Mahidol University in Thailand, and is still published with cooperation of (now) SIL
International. Curiously, we note that at the first ICAAL it had been suggested to change the title of MKS to
Austroasiatic Studies, but this did not follow through, and instead that title was given to the ICAAL
proceedings.
Despite the great success of ICAAL 1, circumstances had begun to turn. The dramatic political
realignment of Indo-China in 1975 saw attention turn away from the region generally, and within the US
(which had funded the first meeting) linguistics was increasingly in the grip of the generative movement
which was scarcely interested in empirical language studies. The existing momentum was sufficient that a
second meeting was held at Central Institute of Indian Languages in Mysore in1978, and attended by a few of
those who had been present 5 years earlier in Hawaii, plus numerous South Asian scholars. This second
ICAAL was held as a post-plenary session of the International Congress of Anthropological and
Ethnological Sciences. David Stampe and Gerard Diffloth agreed to take on the task of editing the
proceedings, but unfortunately that project was abandoned after various delays. As luck would have it, a
bound set of mimeographed papers from the meeting is kept at the Central Institute of Indian Languages in
Mysore, and a mostly readable photocopy made its way into the Cambridge University library. Versions of
several of the papers did make it into press as journal articles or book chapers, but a gap remains in the
history of ICAAL.
Preface v
Subsequently there were no independent international Austroasiatic meetings. There were various
special sessions at conferences devoted to other Asian language families (such as the annual International
Conferences on Sino-Tibetan Languages and Linguistics) and this ad hoc arrangement served the scholarly
community during the lean years of the 1980s and 1990s. But with the new millennium, and the resurgence
of language documentation and emerging interdisciplinary interest from archaeology, genetics etc. the need
for a regular Austroasiatic conference became imperative. This was recognized by Grard Diffloth, Michel
Ferlus and George van Driem in a discussion they held in 2001, and later Diffloth and van Driem were able
to organize an informal gathering, which they called the ICAAL 3 Pilot Picnic, at Siem Reap in June 2006
(with support from Dutch national research council (NOW) and lcole franaise dextrme-orient).
The Pilot Picnic was such a success that it paved the way for the rebirth of ICAAL with a full blown
conference (ICAAL 3) which was held at Deccan College Post-Graduate & Research Institute in Pune,
November 26-28, 2007. The principal organizer was Professor Keralapura Shreevinasaiah Nagaraja, and the
meeting was also supported by the Central Institute of Indian Languages in Mysore. The proceedings have
been edited for publication by Prof. Nagaraja, and will be published as Austro-Asiatic Linguistics: In memory
of R. Elangaiyan (CIIL Publications).
ICAAL 4
In 2009 ICAAL 4 was held at the Research Institute for Language and Culture of Asia, Salaya
campus of Mahidol University (Thailand) 29-30 October 2009. There were some 92 registered participants,
delivering 70 papers over two rather long days (the full program can be seen online at http://icaal.org/).
Reflecting the meeting theme of An Austroasiatic Family Reunion participants came from a wide range of
Asian countries including Thailand, Malaysia, Vietnam, Laos, Myanmar, India, Bangladesh, Nepal,
Singapore and China, as well as western nations.
Participants from the host institution made an especially significant contribution discussing projects
which are documenting and revitalizing endangered languages of Thailand, under the leadership of Prof.
Suwilai Premsrirat (ret.), who has worked tirelessly for Thailand minorities during her distinguished career.
Comparative and historical papers were also prominent, since there is increasing attention to the role that
linguistics can play in discovering the prehistory of Southeast Asia. Papers hoping to discover the
Austroasiatic homeland, and other historical questions, were delivered by George van Driem (Netherlands),
Paul Sidwell (Australia), Michel Ferlus (France), Roger Blench (Britain), Gerard Diffloth (Cambodia). In
connection with these there was also discussion of the application of new statistical methods in language
classification (methods originally developed for genetics), with papers read by Russel Gray (New Zealand),
Jerry Edmondson (USA) and Kenneth Gregerson (USA). Generally it was the most important meeting of
historical linguists on these questions since the first first ICAAL meeting in 1973.
Very important was the participation of scholars from South Asia, made possible by financial
support provided the Centre for Research in Computational Linguistics (Bangkok), which was also a partner
in organising and running the meeting. All up there were nine papers concerning the Munda, Khasi and
Nicobar languages spoken in India and elsewhere in South Asia. This marked a high point in realising the
conference theme, intended to bring scholars together who might otherwise never meet.
The Keynote address was delivered by retired Professor of Southeast Asian Linguistics at Cornell,
Franklin Huffman. Prof. Huffman worked in Southeast Asia during the 1960s and 1970s, and among his
achievements were the development of extensive teaching materials, bibliographic resources, and a widely
used dictionary of Cambodian (for example, it used by the United Nations Transitional Authority in
Cambodia 1992-93). So important was this work, that some have credited him with saving the Cambodian
language after the chaotic Khmer-Rouge years destroyed so much educational and cultural infrastructure.
A special highlight of the meeting, marking the real continuity of the movement over four decades,
was a special presentation of gifts honoring senior figures in the field, including several who had been
present at the 1973 Hawaii meeting. Honored were: Geoffrey Benjamin, Theraphan L-Thongkum, Gerard
Diffloth , Suwilai Premsrirat, Franklin Huffman, and Michel Ferlus.
vi Preface
At the business meeting on the final day participants resolved to launch a new project, a major
handbook of Mon-Khmer languages with chapters on history and typology, and sketch grammars of
representative languages. Mathias Jenny was nominated as chief editor, and an initial group of contributors
nominated. It is expected that the handbook will take several years to produce, and ultimately become one of
the canonical texts of the field.
Finally, the need for an international committee of scholars to ensure the continuity and standards of
ICAAL was discussed, and the following names were nominated and accepted by consensus: Arun Gosh
(India) Brian Migliazza (USA) David Stampe (USA) Doug Cooper (Thailand) Gerard Diffloth (Cambodia)
Jerry Edmondson (USA) Mathias Jenny (Switzerland) Patricia Donegan (USA) Paul Sidwell (Australia)
Roger Blench (UK) Sophana Srichampa (Thailand). Members of this ICAAL International Committee
assisted with review of papers in this volume (in addition to various external reviews) and assist with
assessing and ranking abstract submitted to the conference.
This volume
The present volume begins with the MKS editorial board accepting the invitation of the ICAAL 4
business meeting to produce the ICAAL proceedings as a series of MKS special volumes, with papers subject
to the usual MKS review processes. Principle responsibility for editing and manage the review process fell to
Sophana Srichampa and Paul Sidwell, and Kirk Person also provided special assistance with copyediting and
reviewing. Staff of Mahidol University provided administrative and typesetting assistance, especially Suttilak
Soonghangwa, who deserves special mention. All papers went through a review process, and we thank those
anonymous readers who contributed so much to helping contributors with their paper.
In order to emphasise the continuity of the ICAAL movement we decided to retain the Austroasiatic
Studies name as part of the title. Our intention is to produce these as an ongoing series in a timely fashion,
keeping pace the conference meetings, which are now being held every 2 years. And a note of clarification
concerning the numbering in the sub-title; MKS launched a monograph series in 2000 with Ken Smiths
Sedang Dictionary as the first of these special issues. Others were planned, but did not come to fruition, so
this seems appropriate to breath new life into the series with the ICAAL proceedings. MKS also published
another (self-contained) series, being Aj. Suwilais five volume Khmu dialect dictionaries.
It is our dear hope that this new momentum, reflected in the success of the resurgent ICAAL
meetings, will be maintained, and that the new vigour surrounding the study of Austroasiatic languages will
continue to grow.

Dr. Paul Sidwell
Assoc. Prof. Sophana Srichampa
Bangkok, June 2011

Bedell, George. 2011. The Syntax of Khasi Adverbial Clauses. In Sophana Srichampa and Paul Sidwell (eds.)
Austroasiatic Studies: papers from ICAAL4. Mon-Khmer Studies Journal Special Issue No. 2. Dallas, SIL International;
Salaya, Mahidol University; Canberra, Pacific Linguistics. pp.1-10.
Copyright vested in the author.
1
The Syntax of Khasi Adverbial Clauses
George Bedell
Payap University
Introduction.
Khasi is a Mon-Khmer language spoken primarily in eastern Meghalaya State, India. According to
Ethnologue (Lewis 2009) the number of speakers in India is 865,000; there is considerable internal diversity
however and it is likely that Khasi includes more than one language. The Khasi understood in this paper is
Standard Khasi as used for literary purposes, and recorded in Roberts (1891), Nissor Singh (1904) and the
standard Khasi translation of the Christian Bible. The examples cited are taken from Ka Khubor jong ka
Jingieit (2000), and given in the orthography used there, unless otherwise noted. The numbers indicate
chapter and verse in Ka Gospel U Mathaios (The Gospel According to Matthew). This translation was made
in the nineteenth century and is sometimes difficult for modern speakers to understand; it remains in common
use.
Subordinate Clauses.
Subordinate clauses in Khasi are most often marked with the particle ba, either alone or with another
preceding particle prefixed. Clauses headed with ba may have a variety of interpretations: Roberts (1891;
244, pp. 205-6) mentions three. Sentence (1) illustrates a complement clause, and sentences (2) and (3),
adverbial clauses.
(1) U Kpa jong phi uba ha bneng u tip
3SM father of youPL 3SM=C in sky 3SM know
[ba phi dokam ia kine kiei kiei baroh].
C 2PL need OBJ 3PL=this 3PL=Q 3PL=Q all
your Father in heaven knows [(that) you need all these things] (6:32)
(2) ngan leh ia kaba bha aiu, [ba
1S=FUT do OBJ 3SF=C good what C
ngan ioh ia ka jingim bymjukut]?
1S=FUT get OBJ 3SF ACT=alive endless
what good thing should I do [(in order) to get eternal life]? (19:16)
(3) Phi bakla, [ba phim tip ia ki jingthoh,
2PL wrong C 2PL=NEG know OBJ 3PL ACT=write
lymne ia ka bor U Blei].
or OBJ 3SF power 3SM God
you are wrong, [because you do not know the scriptures or Gods power] (22:29)
There are also relative clauses, exemplified by (4).
2 The Syntax of Khasi Adverbial Clauses
(4) ki khynnah [ba ki iapyrta ha ka templ],
3PL child C 3PL COLL=shout in 3SF temple
the children [(who are) shouting in the temple] (21:15)
A complement clause as in (1) is one which serves as an argument of a verb (here tip know). Adverbial
clauses as in (2) and (3) modify a verb (leh do in (2)) or the main clause (phi bakla you are wrong in (3)).
A relative clause as in (4) is one which modifies a noun (here khynnah children).
It is generally the case that the complement of ba can serve as an independent main clause;
corresponding to (1) to (4) are (5) to (8).
(5) phi donkam ia kine kiei kiei baroh
2PL need OBJ 3PL =this 3PL=q 3PL=q all
you need all these things
(6) ngan ioh ia ka jingim bymjukut
1S=FUT get OBJ 3SF ACT=alive endless
I will get eternal life
(7) phim tip ia ki jingthoh, lymne ia
2PL=NEG know OBJ 3PL ACT=write or OBJ
ka bor U Blei
3SF power 3SM God
you do not know the scriptures or Gods power
(8) ki iapyrta ha ka templ
3PL COLL=shout in 3SF temple
they are shouting in the temple
The initial phi in (5) and (7), nga in (6) and ki in (8) are markers of subject-verb agreement and characteristic
of independent clauses. The future tense -n is suffixed to the agreement marker in (6), and the negative -m is
suffixed to it in (7). See Bedell (2008) for discussion of this agreement.
Structures 1.
As shown in (i), (ii) and (iv), we take ba to belong to the category C (conjunction or complementizer)
which combines with a complement clause (AgP).

The resulting CP may occupy an argument position as in (i).
(i) AgP
NP
U Kpa jong phi
uba ha bneng
Ag'
u VP
CP tip
ba
phi donkam ia kine
kiei kiei baroh
AgP
George Bedell 3

Its syntactic position in (ii) is similar but there it is not an argument.

Its position in (iv) is quite distinct, attached to NP rather than VP.
Adverbial Clauses.
(9) to (11) further illustrate Khasi adverbial clauses; each of them consists of a well formed clause as
in (12) to (14) preceded by a conjunction haba when, naba for or kumba as.
(9) haba u mynsiem bymkhuid u la mih noh
when 3SM spirit unclean 3SM PAST go out
na u briew,
from 3SM person
when an unclean spirit has gone out of a man (12:43)
(10) naba u hap jyndei ha ka ding, jyndei ha
For he fall much into 3SF fire much into
ka um ruh.
3SF water also
for he falls often into the fire and often into the water (17:15)
(11) kumba U Jonah u la shong lai sngi bad lai
as 3SM Jonah 3SM PAST sit three day and three
miet ha ka kpoh u dohkha khraw,
night in 3SF belly 3SM fish big
as Jonah spent three days and nights in the belly of the whale (12:40)
(12) u mynsiem bymkhuid u la mih noh na u riew
3SM spirit unclean 3SM PAST go out from 3SM person
an unclean spirit went out of a man
(ii) AgP
TP
VP
CP
nga
n
leh
ba ia kaba
bha aiu
ngan ioh ia
ka jingim
bymjukut
PP
AgP
(iv) NP
ki khynnah CP
ba AgP
ki VP
iapyrta ha ka templ
(13) u hap jyndei ha ka ding, jyndei ha ka um ruh
he fall much into 3SF fire much into 3SF water also
he falls often into the fire and often into the water
(14) U Jonah u la shong lai sngi bad lai miet
3SM Jonah 3SM PAST sit three day and three night
ha ka kpoh u dohkha khraw,
in 3SF belly 3SM fish big
Jonah spent three days and nights in the belly of the whale
Roberts (1891) lists naba and kumba among copulative conjunctions (p. 122). Though he briefly discusses
the syntax of conjunctions (pp. 203-209), he says nothing about the internal structure of clauses containing
them. Rabel (1961) also lists them as conjunctions, noting that they are compounds of a preposition and ba
(p. 72). She does not discuss their syntax. Nagaraja (1985) discusses them under ba clauses (8.2.8, p. 97).
They are composed of a preposition or prepositional adverbial followed by ba, but it is unclear what syntactic
category ba or its compounds belong to. They are apparently not conjunctions.
Structures 2.
As shown in (ix), (x) and (xi), we take haba, naba and kumba to belong to C, parallel to ba in (ii).
The complement clauses in (9) and (11), unlike (2) and (12), contain overt subjects.

(ix) CP
AgP
Ag' NP
u mynsiem
bymkhuid
TP
VP
u
la
haba
PP mih noh
na u briew
(x) CP
AgP
VP u
naba
PP hap jyndei
ha ka ding
George Bedell 5

Prepositions.
Haba in (9) appears to consist of the preposition ha in, into combined with ba; naba in (10)
appears to consist of the preposition na from with ba, and kumba in (11) appears to consist of the
preposition kum like with ba. These prepositions are illustrated in (15) through (20).
(15) na ka um
from 3SF water
out of the water (3:16)
(16) na u briew
from 3SM person
out of a man
(17) ha ka ding
into 3SF fire
into the fire
(18) ha uta u briew
to this 3SM person
to this man (12:13)
(19) kum u thiat,
like 3SM yeast
like yeast (13:33)
(20) kum u briew uba u trai ing,
like 3SM person 3SM-C 3SM owner house
like the owner of a house (13:52)
Note that (16) and (17) appear as constituents of (10) and (9).
Structures 3.
We take prepositions to be heads of prepositional phrases which take a complement NP as in (xvi),
(xvii) and (xx).

TP
la
(xi) CP
Ag'
VP
U Jonah
kumba
shong lai sngi bad
lai miet ha ka kpoh
u dohkha khraw
NP
AgP
u
PP
u briew
(xvi)
na NP

Note that the conjunctions haba, naba and kumba belong to a distinct syntactic category from the related
prepositions ha, na and kum. Though clearly similar, the meanings do not correspond exactly. As
conjunctions, haba and naba lack any directional component.
Deictics.
In addition to the conjunctions haba, naba and kumba, the prepositions ha, na and kum also appear
attached to deictic stems -ta that, -ne this and interrogative -no, as in (21) through (29). For discssion of
Khasi deictics, see Bedell 2009c, and of Khasi interrogatives, Bedell 2009b.
(21) u la don hangta weibriew.
3SM PAST be there one-person
he stayed there alone (14:23)
(22) um don hangne;
3SM=NEG be here
he is not here (28: 6)
(23) hangno ba yn kha ia U Khrist.
Where C FUT give-birth OBJ 3SM Christ
where the Christ will be born (2: 4)
(24) u la leit noh nangta.
3SM PAST go away from-there
he went away from there (19:15)
(25) To leit noh nangne hangtai,
IMP go away from-here there
go away from here to over there (17:20)
(26) nangno te ka la ioh ki iut?
from-Q and 3SF PAST get 3PL weed
and where did the weeds come from? (13:27)
(27) uban hikai kumta ia ki briew,
3SM-C=FUT teach like-that OBJ 3PL person
who teaches people like that (5:19)
(28) phin duwai kumne:
2PL=FUT pray like-this
you will pray like this (6: 9)
PP
ka ding
(xvii)
ha NP
PP
u briew
(xx)
kum NP
CP
NP
u trai ing ubai
NPi
George Bedell 7
(29) kumno ba u leit hapoh ka templ,
like-Q C 3SM go into 3SF temple
how he went into the temple (12: 4)
Roberts (1891) lists hangne and nangta as adverbs of place (p. 117) and kumno as an adverb of manner(p.
114). They are not mentioned in his discussion of adverbial syntax. Rabel (1961) analyzes them as pronouns
with a locative base (hang-, nang-) or preposition (kum) followed by a demonstrative (-ne, -ta) or
interrogative (-no) base (pp. 67-69). Nagaraja (1985) analyzes them as adverbials, but otherwise follows
Rabel (4.4 and 4.5, pp. 44-47).
Structures 4.
In structures like (xxiii), (xxiv) and (xviii), we take the combination of a preposition and a deictic or
interrogative stem to be a compound preposition which takes no complement. The prepositional properties
are retained in these compounds unlike the compound conjunctions.

Agreement.
Haba, kaba and kumba contrast with ha kaba in (30), na kiba in (31) and kum kiba in (32). These
are not adverbial clauses at all, but rather relative clauses without an overt head noun. The ka- or ki- which
intervenes between the preposition and ba is the agreement characteristic of Khasi relative clauses. In most
such clauses it is optional (as in (4) above), but appears to be required after a preposition. For discussion of
Khasi relative clauses see Bedell 2009a.
CP
u Khrist
(xxiii)
hangno
PP C'
TP ba
yn VP
PP
NP
kha
ia
(xxiv)
nangta
AgP
TP u
la VP
PP leit noh
(xxviii)
kumne
AgP
TP phi
n VP
PP duwai
(30) ha kaba long ka spah jong phi, (6:21)
at 3SF-C be 3SF riches of youPL
where your riches are
(31) shibun na kiba long shuwa (19:30)
many from 3PL-C be first
many who are first
(32) kum kiba iap.
Like 3PL-C die
like dead people (28: 4)
Neither Roberts (1891) nor Rabel (1961) appear to recognize ha kaba or na kiba as a distinct construction.
Nagaraja (1985) regards them as like haba or naba, but with an additional pronominal marker (8.2.8, p. 98).
Structures 5.
Note the contrast between the relative clause structure in (xxx) and the embedded question structure
in (xxiii), both expressed in English by where which may be an interrogative or relative pronoun.

The Khasi adverbal clause type in (9), (10) and (11), though common, does not exhaust the possibilities.
There is, for example, a temporal conjunction mynba when as in (33).
(33) mynba u la thngan,
when 3SM PAST be-hungry
when he was hungry (12: 3)
(xxx)
long ka spah
jong ki
CP
NPi ha
VP
PP
kaba
NPi
(xxxi)
long shuwa
CP
NPi na
VP
PP
kaba
NPi
(xxxii)
iap
CP
NP i kum
VP
PP
kiba
NP i
George Bedell 9
There is an associated deictic as in (34) and interrogative as in (35), but myn is not used as a preposition.
(34) uba mynta u long,
3sm=c now 3sm be
which now exists (6:30)
(35) mynno ba ngi la iohi ba me la thngan, ...?
when C 1PL PAST see C 2SM PAST be-hungry
when did we see that you were hungry? (25:37)
There are also many prepositions which do not form conjunctions as well as conjunctions which are not
related to a preposition. An example of the latter is lada if.
(36) lada un sngew ia me,
if 3SM=FUT listen OBJ youSM
if he listens to you (18:15)
Abbreviations
1S first person singular
1PL first person plural
2SM second person masculine singular
2SF second person feminine singular
2PL second person plural
3D third person diminutive
3SM third person masculine singular
3SF third person feminine singular
3PL third person plural
ACT action nominalizer
AGT agent nominalizer
C complementizer
CAUS causativizer
CLASS classifier
COLL collective
DIM diminutive
EMPH emphatic
FUT future tense
IMP imperative
LOC locative
NEG negative
OBJ object marker
Q interrogative
PAST past tense
PROX proximate
SUBJ subjunctive
VOC vocative
youSM you (singular masculine)
youSF you (singular feminine)
youPL you (plural)
References
H. Roberts, A Grammar of the Khassi Language, London: Kegan Paul, 1891 (reprint New Delhi: Mittal
1995).
U Nissor Singh, Khasi-English Dictionary, Shillong, 1904 (reprint New Delhi: Gyan 2003).
L. Rabel, Khasi, a Language of Assam, Baton Rouge: LSU Press, 1961.
K. S. Nagaraja, Khasi, a Descriptive Analysis, Pune: Deccan College Post-Graduate & Research Institute,
1985.
Ka Khubor jong ka Jingieit (The Message of Love: the New Testament in Khasi), Bangalore: Ka Bible
Society jong ka Ri India, 2000.
G. Bedell, The Syntax of Agreement in Khasi, presented to NEILS 3, 2008.
G. Bedell, Agreement in Khasi Relative Clauses, presented to NEILS 4, 2009a.
G, Bedell, The Syntax of Khasi Questions, presented to NEHU International Conference, 2009b.
G. Bedell, The Syntax of Deixis in Khasi, presented to SEALS 19, 2009c.
Lewis, M. Paul (ed.), Ethnologue: Languages of the World, Sixteenth edition. Dallas, Tex.: SIL International,
2009. Online version: http://www.ethnologue.com/

Benjamin, Geoffrey. 2011. Deponent Verbs and Middle-Voice Nouns in Temiar. In Sophana Srichampa and Paul
Sidwell (eds.) Austroasiatic Studies: papers from ICAAL4. Mon-Khmer Studies Journal Special Issue No. 2. Dallas, SIL
International; Salaya, Mahidol University; Canberra, Pacific Linguistics. pp.11-37.
11
Deponent Verbs and Middle-Voice Nouns in Temiar
Geoffrey Benjamin
Nanyang University
Abstract
The infix -a- in Temiar (Central Aslian, Peninsular Malaysia) covers a wide range of
functions, both productive and non-productive. In Temiar morphology generally it serves
iconically to indicate an embedded object. As a productive infix, it forms the middle voice
of verbs. The main non-productive function of -a- is as a frozen infix in (1) a set of non-
inflecting deponent verbs and (2) a set of disyllabic (middle-voice) nouns. The same
semantic dimension underlies both, namely that the verbs and nouns in question indicate
that the subject or entity is thought of as being simultaneously its own agent and patient.
Culturally, the special attention paid to middle-voice processes relates to a central Temiar
interest in the specifically dialectical character of SelfOther relations.
1. Introduction
Temiar, currently spoken as a first language by approximately 27,000 people in northern Peninsular
Malaysia, belongs to the Central Aslian (Senoic) division of Mon-Khmer (Diffloth 1975; Benjamin 1976a,
2004; Matisoff 2003). It is also spoken by a few thousand other people as a lingua franca among the northern
Peninsular Orang Asli (Aborigines) and some Malays. Based on my own summary account of Temiar
grammar (Benjamin 1976b), a moderately extensive secondary literature has emerged, concerned mostly with
the rich reduplicative morphology exhibited by the Temiar verb (Tables 1 and 2).
1
However, the majority of
those analyses are based on an approach to the Temiar verb that I later revised. In my 1976 account I treated
the verbal infix -a- as indicating what I called the simulfactive mode. I now regard this infix as indicating,
not mode (or Aktionsart), but voice (or diathesis), i.e. participant orientation. Specifically, the infix -a-
indicates that the verb is in the middle voice, as part of a series that also includes an uninflected base form as
well as an inflected causative in -r-.
2

1.1 Inflecting verbs
Tables 1 and 2 illustrate the forms taken by the two different morphological categories of inflecting
verbs: those with a monosyllabic base and those with a sesquisyllabic base. Each form is given twice, first in

1
The secondary studies include McCarthy (1982: 208221), Broselow & McCarthy (1983: 3843), Ter Mors
(1983: 280288), Sloan (1988: 319324), It (1989: 252254), Spencer (1991:151, 155), Carstairs-McCarthy (1992:84
85), Shaw (1994: 122127), Spaelti (1997: 112, 134135, 162165), Gafos (1998a: 233249, 1998b: 517522, 1999: 75
107), Walther (2000), Raimy (2000: 145151), Hendricks (2001: 290), Idsardi & Raimy (n.d), Frampton (2004:160163)
and Yap (2009).
2
I first presented this revised paradigm in Benjamin 2001: 114. So far, the only secondary accounts to have
incorporated it are Matisoff 2003 and Yap 2009. A fuller discussion of Temiar morphology is presented in Benjamin
2011; the present paper derives from an earlier draft of that study. Commentators have employed several different labels
for the reduplicative pattern exhibited by the imperfective forms in Tables 1 and 2. My preference is for Matisoffs self-
explanatory iconic term incopyfixation (Matisoff 2003: 2832).
12 Deponent Verbs and Middle-Voice Nouns in Temiar

a strictly phonemic orthography and second in an orthography identical to my own practical orthography for
Temiar. The phonemic orthography indicates the morphology more clearly, but is harder to read than the
practical orthography. (Except where otherwise noted, I employ the practical orthography in the main body
of this paper.) The glosses indicate some of the semantic distinctions signaled by the various forms.
Table 1: Monosyllabic: gl to sit

VOICE ASPECT VERBAL NOUN
Perfective Imperfective Progressive
Base slg [slg]
sleep (completed
act)
sglg [sglg]
sleep (incomplete act)
bar-sglg [bar-sglg ~
b-sglg] currently
sleeping
snlg [snlg];
snglg [snglg]
a sleeping
Middle salg [salg]
fall asleep (uncontrolledly)
bar-salg [bar-salg ~
b-salg]
currently falling asleep;
sleep
(with someone)
snalg [snalg]
a falling asleep
Causative srlg [srlg]
put (someone)
to sleep
(completed act)
srglg [srglg]
put (someone)
to sleep (incomplete act)
bar-srglg [bar-srglg ~
ba-srglg]
currently putting (someone)
to sleep
srnlg [srnlg]
a putting
(someone)
to sleep
Table 2: Sesquisyllabic: slg to sleep (also: lie down, marry)

The imperfective aspect is formed by copying the verbs final consonant as an infix, leftwards,
before the final syllable. The middle voice is formed by infixing -a- into the same position. With
monosyllabic verbs (Table 1) the initial consonant is copied as well, in both the imperfective and the middle
forms. (For the morphophonemic changes and morphological processes involved, see Benjamin 1976b: 143
145, 168170.)

3
During my initial period of fieldwork in the 1960s and 1970s verbal nouns with prefixed word-initial n- were
heard only in the state of Perak. More recently, this form is also being heard across the watershed in Kelantan, alongside
the older infixal form. Could this indicate an increasing intolerance of infixes by people living in more modern
circumstances?
Base gl [gl]
sit (completed act)
glgl [glgl]
sit (incomplete act)
bar-glgl [bar-glgl ~
b-glgl]
currently sitting
glnl [glnl]
~ nlgl [nlgl]
a sitting
3

Middle gagl [gagl]
sit (uncontrolledly,
or all together at once)
bar-gagl [bar-gagl ~
b-gagl]
currently seated
(uncontrolledly)
gnagl [gnagl]
a sitting
(uncontrolledly)
Causative trgl [trgl]
make (someone) sit
(completed act)
trlgl [trlgl]
make (someone) sit
(incomplete act)
bar-trlgl
[bar-trlgl ~
ba-trlgl]
currently making (someone)
sit
trngl [trngl]
a making
(someone) sit
Geoffrey Benjamin 13

1.2 Non-inflecting verbs
Temiar also possesses many disyllabic verbs that exhibit almost none of the inflectional possibilities
displayed by the monosyllabic and sesquisyllabic verbs.
4
Their sole inflection is nominalization with -n-; but
they also form a progressive-cum-imperfective with the proclitic bar- ~ b-.
5
As Tables 3 and 4 illustrate,
these fall into two subtypes: those with -a- in the first syllable (Table 3) and those with some other non-
predictable vowel, such as -o- or -e- (Table 4).
6
For reasons explained later, I shall refer to most of the verbs
with -a- in the first syllable as deponent verbs.
7

ASPECT VERBAL NOUN
Perfective Imperfective, Progressive
halab [halab]
travel downriver
bar-halab
[bar-halab ~ b-halab]
currently travelling downriver
hnalab [hnalab]
a journey downriver
Table 3: Deponent: halab to travel downriver

ASPECT VERBAL NOUN
Perfective Imperfective, Progressive
golap [golap]
carry on shoulder
bar-golap
[bar-golap ~ b-golap]
currently carrying on shoulder
gnolap [gnolap]
a carrying on shoulder
Table 4: Disyllabic: golap to carry on ones shoulder
2. The Temiar middle voice
Before proceeding to the primary topic of this paper deponent verbs and middle-voice nouns let
me first outline the normal productive usage of the middle voice in Temiar. Temiar is the only Aslian
language that makes use of -a- as a fully productive affix in the inflection of the verb; its close relative Lanoh
also employs this affix, but less productively. For the reasons outlined below, it seems clear that middle

4
These include a moderate number of verbs with more than three consonants, such as sindul to float, or verbs
with a further presyllable, such as smaar to snore.
5
As indicated in Tables 1 and 2, the progressive clitic bar- is almost universally attached to the reduplicated
imperfective form of the verb. However, I recorded one or two cases in which bar- was attached instead to the
unreduplicated perfective form of the verb, as in w-b-kooy k
[2DU-PROG-roast.PFV fish] they were roasting the fish

(instead of the expected w-b-kykooy k
). Since these instances were rare, it is impossible to determine whether they

were intentional or simply speakers mistakes.
The following glossing conventions and abbreviations are employed in this paper: * non-occurring or
hypothetical form; - (hyphen) proclitic; < > infix; 1 1st person; 2 2nd person; 3 3rd person; ACC accusative; C
i

initial consonant; DU dual; EXPR expressive; INCL inclusive; INT intentive; IPFV imperfective; MID middle voice;
NOM nominative; PFV perfective; PL plural; PROG progressive; SG singular; VET vetative (negative imperative).
6
In my earlier study (Benjamin 1976b: 138), I described the minor-syllable vowel as containing a single
phoneme, //, with a variety of allophonic pronunciations. It would have been better simply to describe the various
pronunciations of what is, in fact, a non-phonemic epenthetic transition. These pronunciations vary between [], [i] and
zero, depending on the following consonant and the vowel in the word-final major syllable. In a strictly phonemic
transcription, therefore, these minor vowels should be omitted, but it is still useful to incorporate them into a practical
orthography. In the past, I have tended to write them all as , but in future I shall write i instead, where appropriate: sik
pandanus instead of sk (phonemically /sk/). In the majority of Temiar dialects, the predictable non-phonemic
epenthetic vowel is regularly pronounced [] in closed syllables (i.e., before -CCV-), as written in my practical
orthography.
7
As noted in Section 3.2 and the Appendix below, some verbs in -a- do not qualify as deponent, and should
preferably be placed in the disyllabic category.

voice best describes the character of these -a-inflected forms. Simulfactive, the term I employed previously
(Benjamin 1976b: 172174), refers to just one of the uses of the middle voice. I discuss it below under the
label of all-together middle.
The infix -a-, which also occurs in a range of non-verbal contexts, signals the notional meanings
OBJECT or SALIENT OTHER. I hold that, in doing so, it reflects the iconic effects of contrasting deictic gestures
made by the vocal organs:
Opening the mouth wide, to address oneself to the rest of the world
Closing the mouth in self-contemplation, as if in temporary retreat from the world.

The relatively open mouth position, which signifies the directing of ones attention to the objective
realm of OTHER (a generalization, perhaps, of you [h]-deixis), is expressed phonetically in Temiar by the
low vowel a, the back consonants and h, and vowel nasality (i.e., velic opening). The relatively closed
mouth position, which conversely signifies the more subjective SELF-focused, I-deixis realm, is expressed
phonetically by the high vowel i and the front consonants m, y, c and r.
8

Elsewhere (Benjamin 2011), I have argued that this iconic pattern is manifested in a significant
degree of phonetic and morphological syncretism in Temiar between (1) pronouns, (2) demonstratives
(deictic particles), (3) definitizing clitics, (4) the inflection of human nouns for number, (5) the marking of
the sentence for mood and orientation, (6) the marking of discourse continuity and (7) the inflections of the
verb for voice and aspect. However, there is no space here to discuss these other features. The present paper
is concerned only with the morphology of certain verb forms as well as related features displayed by some
nouns.
As displayed in Tables 1 and 2, the middle voice of the Temiar verb is formed by inserting the
OTHER-referring infix -a- into the presyllable This carries with it several interconnected semantic
implications. First, it indicates that the middle voice is regarded as object-incorporating, and hence heavily
intransitive.
9
Second, these features imply in turn that the middle voice refers primarily to dynamic situations,
and to events rather than processes. Third, intransitive actions or processes not proceeding beyond the self are
more likely to be thought of as punctiliar than actions that bring about a change in an external object, which
take more time and hence are more likely to be durative or iterative. Thus the employment of a verb in the
middle voice carries with it the implication that the situation it refers to is a DETERMINATE EVENT that has
become an OBJECT of attention in its own right.
This prototypically punctiliar quality of the middle voice explains the asymmetry in Tables 1 and 2,
where only the base-form and causative of the verb possess distinctive inflections for the perfective and
imperfective aspects. Given its peculiar semantic implications, therefore, the Temiar middle voice simply
does not require a special imperfective-aspect inflection.
10
I say protoypically punctiliar because the
semantic features under discussion here are connected by strong tendencies elective affinities, so to speak
(following Max Weber) rather than strict rules of co-occurrence. The features are therefore graded in
character rather than discrete. For example, non-punctiliar usages of the middle voice do sometimes occur, as
in the progressive forms with bar- mentioned below.

8
For a detailed description of the Temiar sound-system see Benjamin 1976b: 130153. For further discussion
of iconicity in Temiar and elsewhere, see Benjamin 2011. Note that the iconicity under discussion here relates to oral-
articulatory gesture rather than the acoustic properties of the sounds produced by those gestures (cf. Jakobson & Waugh
1979: 182).
9
Although I am keeping to the traditional term middle voice (having studied a little Classical Greek in
secondary school), the more recent term unaccusative (Levin & Hovav 1995) often serves as a better characterization of
these usages. Other authors, such as Comrie (1981) use anti-causative for the same set of features.
10
An appropriate morphological pattern could nevertheless be constructed for an imperfective-middle inflection:
gl glgl [glgl] *glagl [*glagl] sit; slg sglg [sglg] *sgalg [*sgalg] sleep. Although these
patterns are not found as regular inflections of the Temiar verb, they are often employed in the highly iconic lexical class
known as expressives, widely employed by speakers of Temiar and other Aslian languages (cf. Diffloth 1972, 1976b,
Benjamin 1976b: 177178).

Given these semantic underpinnings, there are at least three circumstances in which Temiar speakers
might choose to employ the middle voice:
11

The action referred to by the verb is performed with reference to some salient other.
The object of the action or process is incorporated within the verb as part of its meaning. It is this
pattern that is sometimes referred to as unaccusativity (Levin & Hovav 1995).
The action or event referred to by the verb is a particular determinate event treated as a noteworthy
object of comment.
2.1 Salient-other middles
The middle voice is frequently employed when the verb is regarded as involving two or more actors
directing their actions simultaneously to each other or to some common focus. There are at least two such
usages: all-together and reciprocal constructions. In these constructions, the actors treat each other as
SALIENT OTHERS, as marked by the -a- infix of the verb.
In the following examples, the all-together middle-voice constructions (1a) and (1c) are contrasted
with their non-middle equivalents, (1b) and (1d):
a. un-waw. (1)
3PL-depart.MID.
They all departed together.

un-w. b.
3PL-depart.PFV.
They departed.

un-b-salg. c.
3PL-PROG-sleep.MID.
They are all sleeping together.
[Said of a group of cats and kittens.]

un-b-sglg. d.
3PL-PROG-sleep.IPFV.
They are sleeping.

In the following example (2), the first appearance of the verb tap to plant is in the all-together
middle-voice form (tatap), but in its second appearance it is in the non-middle perfective form (tap). (This
example, like many of the others in this paper, is taken from my collection of stories recorded in the 1960s.)
un-b-tatap slaay w-na. un-tap naar is. (2)
3PL-PROG-plant.MID swidden 2DU-that 3PL-plant.PFV two day.
They were all planting the two-of-thems swidden together. They planted for two days.

All-together verbs correspond to the simulfactive meaning that I ascribed to this form of the verb
in my original account (Benjamin 1976b: 172173). This was supported to some extent by those Temiar-
speakers who explained the -a- inflection by saying that it marked a plurality of actors.
12
Clearly, actions

11
On the middle voice as primarily semantic and only secondarily syntactic, see Watkins 1976: 309. Some
authors, such as Arce-Arenales et al. 1994, distinguish between voice as a syntactic category and diathesis as the
equivalent semantic category.
12
Means (1998: 12) seems to be have been told something similar by her committee of Temiar speakers. She
refers to the infixed -a- form of the verb as the plural, and provides a purportedly plural form for many of the verbs
listed in her dictionary. But these -a- forms occur so frequently with singular subjects in ordinary Temiar speech (as in

carried out in or by social groups are very likely to be both determinate and simulfactive, as well as indicative
of the attention that the individual actors must be paying to each other: hence the bringing together of these
various meanings of -a- in the all-together middle-voice inflection. They are not necessarily punctiliar,
however, as all-together middles often occur with dual or plural subjects attached to the verb via the
progressive clitic bar-, as in (1c) and (3). I suspect therefore that these all-together middles may be
underlain by a more fundamentally reciprocal semantic.
The reciprocal actions of two individuals are expressed by linking a dual-number pronoun to a
middle-inflected verb via the proclitic bar- ~ b- progressive, as in (3a).
13
If the event had been non-
reciprocal, with the two protagonists handing food out to third parties rather than to each other, the verb
would have been in the non-middle imperfective, as in (3b).
Lpas na w-b-aog -lah.
14
(3) a.
After that 2DU-PROG-give.MID 3SG-lah.
After that, they were giving each other [some].

Lpas na w-b-gog -lah. b.
After that 2DU-PROG-give.IPFV 3SG-lah.
After that, they were giving [some, to others].
2.2 Object-incorporating (unaccusative) middles
As illustrated in the specially constructed form-sentences (4a)(4e), the semantic interpretation of a
verb inflected for middle voice depends largely on the properties of its grammatical subject. When the
subject is animate and singular the interpretation will usually be that the verb refers to a body-move (4a, 4b)
or to some emotional response (4c, as contrasted with 4d), either (agentively) reflexive or (non-agentively)
medio-passive. In the absence of a positively agentive marking with the intentive infix -m-, it is not always
possible to tell which meaning is intended. The reason for this ambiguity is not hard to find. To say that the
object of an action is incorporated within the verb (as indicated by the infix -a-) is to imply that the action
(medio-passively, unaccusatively) does not proceed beyond the subjects self. This is much the same as
saying that the action is thought of as having no external source. Middle voice and intentive mode are
therefore unlikely to co-occur, and I have not yet found any examples of such a combination. The
simultaneous combination of agentive planning and non-agentive medio-passive emotion depicted in (4e) is
therefore unlikely ever to be more than a hypothetical example.
Na-gagl. (4) a.
3SG-sit.MID.
He sat.
(What did he do?:) He sat down.
(What happened to him?:) He became seated.

examples (7)(9), below) that the plural analysis cannot be correct regardless of the attempts by my and Meanss
Temiar respondents to explicate their language. Consequently, some of the -a- forms of the verb given in Meanss
dictionary do not actually occur in the language, since they are semantically inappropriate or impossible, while those that
do occur are actually middle-voice, not plural, forms. In Temiar, only adjectives and human nouns inflect for plurality
though not through infixation of -a-, which paradoxically indicates the singular number in human nouns (Benjamin
1976b: 185)!
13
Burenhult (2005: 104) reports that Jahai, a Northern Aslian language, employs the C
i
a-/-a- morphology as a
reciprocal inflection in a few verbs. Although he doesnt say so, this usage was probably borrowed from Temiar, a
language that most Jahais can speak.
14
The phrase -lah is an idiomatic expression marking the declaratory character of the sentence that it
terminates, especially in story-telling. There is no satisfactory way of translating it into English.

Na-salg. b.
3SG-lie.MID.
He lay down.
(Agentive:) He lay down and slept.
(Non-agentive:) He fell asleep.

Na-yayaap. c.
3SG-weep.MID.
She burst uncontrolledly into tears.

Na-yaap. d.
3SG-weep.PFV.
She wept.

*Nam-yayaap. e.
3SG<INT>-weep.MID.
*She intentionally burst uncontrolledly into tears.

An authentic text-example is provided in (5), which tells of the final success of two hungry men in
hunting a bird to eat. The details of the cooking are narrated in a string of straightforward perfective verbs.
But the final drinking of the soup is reported in the emotional middle voice (hahuj, not huj), indicating that
they were so hungry that they couldnt help but drink the soup up, and were therefore acting (medio-
passively) under compulsion.
Gd i-kw
s ha-c
s na, gd yl, (5)

Cut.PFV NOM-child ACC-bird that, cut.PFV finish,
na-pl ma-oos, na-didih k na,
3SG-cook.PFV to-fire, 3SG-boil.PFV water that,
w-koh, w-koh, w-b-hahuj.
3DU-pour.PFV, 3DU-pour.PFV, 3DU-PROG-drink.MID.
The son cut up the bird, finished cutting he cooked it on the fire, the water boiled, they poured
it out and they drank it up.

With an inanimate subject or with a verb that carries an inherently passive or adversative meaning
there is little chance of ambiguity, for then the non-agentive interpretation of the middle voice must apply. It
is this that gives rise to the passive-like absolutive constructions exemplified in (6a), as contrasted with the
agentive, non-middle, employment of the same verb in (6b):
Kb doh na-wawg. (6) a.
fruit this 3SG-open.MID.
This fruit split open.

i-wg kb doh. b.
1SG-open.PFV fruit this.
I opened this fruit.

From a semantic point of view (6a) constitutes a passive, as the grammatical subject governing the
verb (na- 3SG, anaphoric to fruit, the topic) is actually the patient. This spontaneous, non-controlled use
of the middle voice is the nearest thing to a passive construction in Temiar, which possesses no overt passive
inflection.
The employment of the middle voice to indicate absolutive meanings is still productive, as
exemplified by (7). This was said by a Temiar as he stopped his car for a few minutes to allow it to cool

down. He could have employed the simple perfective form geej to burn, but this would have indicated
neither the suddenness nor the apparent absence of external cause expressed by the middle-voice form that he
chose.
i-iid kreth na-gageej. (7)
1SG-fear.that.PFV car 3SG-burn.MID.
I was afraid the car would burn up.

Some verbs almost always occur in the middle-voice form, even though their ordinary perfective
base forms are also occasionally used. Of these, dado flee is an example; the non-middle forms do and
ddo run are also employed, but infrequently. Presumably, fleeing is something that Temiars do under
compulsion, either in terror or mock terror (as in childrens play), rather than as a fully agentive act.
15

2.3 Determinate-happening middles
As already noted, the object-incorporating property of the -a- infix can also be used to indicate an
events noteworthiness or unusual character. Examples of this usage are relatively rare, especially when
unassociated with all-together, reciprocal, absolutive or emotional meanings. One such example is (8).
The middle-voice usage here (lals, instead of the expected ls) expresses my story-tellers affected
surprise that the protagonist would dare to do what he had just warned his wife not to do.
Na-tipuu ma-lh, na-lals, na-pdpood ma-kb t. (8)
3SG-deceive to-wife 3SG-return.MID, 3SG-eat.IPFV to-fruit earlier.
He deceived his wife, and believe it or not! went back and ate the fruit.
[He had earlier (t) forbidden her to eat the fruit, pretending it was poisonous.]

A simpler example, extracted from a conversation that I overheard, is given in (9a); contrast this
with the normal non-middle expression in (9b).
Na-tatd. (9) a.
3SG-stand.MID.
He stood to attention

Na-td. b.
3SG-stand.PFV.
He stood.
3. Deponent (inherently middle) verbs
So far, I have been discussing examples of the productive inflectional use of the verbal infix -a-.
However, as already mentioned, Temiar also possesses a large number of non-inflecting, mostly disyllabic,
verbs. The sole inflection they exhibit is nominalization with -n-.
16

15
Cf. Blagden (1906: 620), entry G44, where various forms of this verb are given in different Temiar dialects:
dd [ddo] and dadu (daduk); dado (dadok), in the phrases run off! em da dok [m-dado PL.INCL<INT>-run.MID!
or perhaps ham-dado 2SG<INT>-run.MID!], and dont run away, ed da dok [probably, a-dado VET-run.MID].
16
This lack of inflection is puzzling. In Semai, the closely related Central Aslian language, such verbs do
exhibit reduplicative inflectional processes. Grard Diffloth has suggested to me that this is because Temiar has had less
time than Semai to absorb these items, which presumably are largely of Austronesian provenance, and from Chamic in
particular. (The verb golap in Table 4 is such an import from Chamic.) My own lexicostatistical data (Benjamin 1976a:
73) confirm that Semai has had more contact with Malay than Temiar has had. Thus, (1) disyllabic verbs, especially those
(like golap) with something other than -a- in the first syllable should be examined for Austronesian etymology, and (2)
forms with -a- in the first syllable should be examined for evidence of having been altered from some other vowel. While
proposal (1) must await another occasion, proposal (2) is mentioned briefly in Section 5.1 below.

Ancient Greek and Latin possessed a class of verbs that carried an apparently active meaning despite
an overtly middle (in Latin grammars, usually labeled passive) morphology. Traditionally, these verbs are
referred to as deponents, because they appear to lay aside, shed (Latin, deponere) the middle or passive
sense suggested by their morphology. Temiar too has such a class of deponents, consisting of a number of
non-inflecting disyllabic verb-stems possessing the full vowel a in the first syllable but with meanings that
appear on initial examination to be wholly active. These verbs lack almost all inflection, so that halab to
travel downriver (Table 3), for example, possesses no perfective *hlab, no imperfective *hblab, and no
causative *hrlab or *hrblab. They do, however, take the progressive clitic bar- ~ b-, which generally
serves with these verbs to express durative or imperfective meanings, as in b-halab travelling downriver.
They also form verbal-nouns by -n- infixation, as in hnalab a journey downriver. Other examples are:
sath to collapse, carh to walk down-hill and gabag to sing.
But do these deponent verbs really possess active meanings? Just as with some Latin deponents
(Baldi 1974: 19), closer examination suggests that these verbs are interpretable as having inherently middle-
voice meanings.
17
For some of the verbs this is not hard to accept. When travelling downriver (halab), one
passively lets ones raft be carried by the current, and the only action the rafter takes is to steer the craft away
from obstructions; the meaning in this case falls into the medio-passive category. When a house collapses
(sath), that is something it undergoes, not something it does: the meaning is essentially the same as that of
the absolutive unaccusative middles discussed earlier. Similarly, going downhill (carh) in Temiar country
is hard for walkers to control, as they descend precipitously through the undergrowth. Carh, then, is
something one suffers to some extent, and the middle-voice morphology is again quite appropriate.
18
In
contrast, the verb tuh go uphill, which undeniably denotes an action, lacks the a vowel. Indeed, I doubt
whether the middle-voice form *tauh would be acceptable to a Temiar speaker, as it would imply an
unacceptable combination of active and passive meanings.
19

But what is one to make of the middle-voice morphology of gabag to sing? Surely, singing is
unambiguously an activity? It is here that culturally derived ideas intervene, for traditional Temiar songs are
not primarily thought of as instances of simple self-expression or public entertainment. Temiars sing either to
quietly entertain a visiting spirit-guide positioned somewhere near their own dream-soul (rwaay), or while
performing ceremonially as mediums through whom visiting spirit-guides sing out their message to the
whole community. In the former case the singer will usually be transiting gradually from dreaming to waking
(typically at daybreak), and the singing will in effect be self-directed. This kind of singing is semantically
like a reflexive body-move (cf. Diffloth 1974), for which the middle-voice inflection if the language
possesses one is the appropriate form of the verb. On the other hand, where the singer performs publicly as
a spirit-medium, he (occasionally, she) will typically be in a light trance and his singing will be conceived of
as a passive experience issuing from a supposedly external agentive source, the spirit-guide. Here again, the
inherently middle-voice morphology of the verb gabag is quite appropriate to the meaning that singing
possesses in Temiar culture.
20
This analysis of the verb gabag both explains and gains support from a puzzling ethnographic
curiosity. During communal song-sessions, Temiars often shout encouragement to the lead-singer with the
phrase ma-rii hah to your self!. One would expect this literally to mean keep it to yourself! (i.e. quieten

17
Diffloth (1976a: 242243) presents a list of disyllabic Semai verbs with -a- in the first syllable. Although
the -a- infix is not productive in Semai (unlike Temiar), many of the meanings there can nevertheless be analyzed as
deponent in character.
18
The likely etymology of carh is instructive: Mon crh to fall, Mon grh to push, shove, fall down in
pieces, crh a to drop water on the ground, Khmer jr (chrh) tomber, se dtacher etc. (Pinnow 1959: 91, no.
106). Temiar therefore has added the -a-, reinterpreting the experience of precipitate downhill movement as something
like to fall and be made to fall. (However, there is also a rarely used Temiar verb crh to chop bamboo down at its
base that could be related to carh.)
19
Conceivably, tauh could be employed for all climbing a hill together, but I have yet to hear this usage.
20
For more on the musical context, see Roseman 1984, who incorporates my analysis of gabag sing on p. 422.
For a directly parallel instance from Malay, see my discussion of the choice between the older middle-voice form
bernyanyi and the now more usual active-voice menyanyi, both meaning sing (Benjamin 1993b: 378). Present-day
Temiars refer to their own and others pop-music singing as aii, borrowing the Malay word.

down!), but the attendant behavior makes it clear that even greater volume and enthusiasm are being called
for. The point, of course, is that the Temiars themselves conceive of public singing-out as a self-directed,
middle activity. Here the dialectical entanglement of SELF and OTHER that is so characteristic of Temiar
culture (see below) infuses both language and behavior simultaneously.
Further examples of middle-voice deponent verbs and adjectives (which in Temiar are stative verbs)
are given in the Appendix below. The listing also gives suggested reasons for the middle-voice semantics
exhibited by these verbs. It may seem strange that many of these deponent verbs are stative, for I earlier
characterized the middle voice as being prototypically dynamic. The answer, it seems to me, is that these
stative deponent verbs represent states that are assumed to be in dynamic equilibrium, held in place by the
simultaneous pull of two opposed forces. Moreover, these verbs are non-inflected lexical items complete in
themselves, rather than inflected middle-voice forms selected from within an inflectional paradigm that offers
such other choices as causative or imperfective.
3.1 Partly-inflecting (semi-deponent) verbs
In the previous sections I discussed only two basic morphological classes of verbs in Temiar:
inflecting and non-inflecting. However, there also exist verbs that fall between these two poles by exhibiting
a restricted range of inflections. These semi-deponent verbs fall into two subtypes: some follow a middle
imperfective distinction (Table 5), while others exhibit a middlecausative distinction (Table 6). In both
cases, the least-inflected form of the verb is a deponent in -a-; there is no simple base form in --.
Base hrk [hrk]
clearing up
(incomplete)
bar-hrk [bar-hrk~
b-hrk]
currently clearing up
hnrk
[hnrk]
a clearing-up
Middle hark [hark]
cleared up (complete)

Causative
Table 5: Partly-inflecting (type 1, basemiddle): hark to clear up (sky, river)

Base
Middle kaleek [kaleek]
to prop, be propped
bar-kaleek
[bar-kaleek~
b-kaleek]
currently propping
knaleek
[knaleek]
a prop
Causative krleek [krleek]
set a prop in place
(completed act)
krleek [krleek]
setting a prop in place
(incomplete act)

Table 6: Partly-inflecting (type 2, middlecausative): kaleek to prop

Typical Temiar props (knaleek, Table 6) are the struts set athwart a dug-out canoe, or the
scaffolding (also knaleek) jammed between the ground and a house or platform to stop it collapsing, or the
bridges on a stringed instrument. What these have in common is that they simultaneously hold two objects
apart while being held in place by those objects. Such props thus affect while being affected which is
exactly what Benveniste (1971: 149) identifies as the chief criterion of the middle voice, as opposed to the

active or passive voice. The verb kaleek therefore deserves its inherently middle-voice morphology; an
active-voice base-form morphology (*kleek) would not be appropriate. The causative krleek, meaning to
set a prop or scaffold in place, is in regular use, but the semantic implications of choosing between this and
the semi-deponent middle voice are not always obvious, as illustrated in (10).
Na-kaleek ya-ha-jh -na, na-kaleek (10)
3SG-scaffold.MID yes-ACC-tree 3SG-that, 3SG-scaffold.MID
gggg blas ma-jalk h.
until [EXPR] all.way to-top 3SG.
He scaffolded the tree, he scaffolded all the way up to its top.

Here, my story-teller chose not to employ the causative krleek, even though that might seem to
have been the appropriate form. Instead, he employed the semi-deponent middle-voice form kaleek,
regardless of the accusatively marked object (ha-jh the tree) that it appears to govern. His aim,
presumably, was to emphasize how the climber was supported himself as he ascended and to downplay what
was being done to the tree. If, on the other hand, the story had required the protagonist to prop up a rickety
house to prevent it collapsing, the narrator would probably have chosen the causative forms krleek or
krleek.
Further examples of such type 2 partly-inflecting verbs are:
catk to close (said of a door closing by itself: absolutive, non-agentive, intransitive), crtk to
close (said of someone closing a door: agentive, transitive)
lak to break off, come off (said of something falling off spontaneously: absolutive, non-agentive,
intransitive), as opposed to lrk to break something off, remove (agentive, transitive).

An overheard example of the latter is given in (11). This reported on the success of a surgeon in
repairing the damage done to a Temiar road-accident victim.
Na-lrk bsii num-gntk. (11)
3SG-come.off.CAUS iron from-ear.
He removed metal from the ear.
3.2 Non-deponent verbs in -a-
As noted earlier, there are several non-inflecting Temiar verbs with -a- in the initial syllable, but
which do not exhibit any clearly middle-voice semantic. (See the Unclassified section in the Appendix.)
Examples are sapood to wrap, make a parcel (an Austronesian loan) and dalag to call (someone). Further
investigation may yet reveal an underlying middle-voice semantic in many of these verbs. But, given that [a]
is the most common vowel in the languages of the world, it is not surprising that this too sometimes occurs in
the presyllable of non-deponent verbs, along with such vowels as [o] or [e].
4. Verbs in -a- and Temiar culture
As we have seen, there are at least four classes of verbs with -a- in their presyllable:
Productive middle-voice forms, as part of a fully inflecting verb paradigm that also contains active
and causative forms: gaglgltrgl sit (Table 1) and salgslgsrlg sleep (Table 2).
True deponents, non-inflecting or partly-inflecting verbs with an underlying middle-voice
semantic: halab travel downriver (Table 3), hark clear up (Table 5).
Partly-inflecting verbs that are agentless (and intransitive) in their primary middle-voice form, but
which also possess an agentive causative-transitive inflection: lak to break off (intrans), lrk to
break off (trans), but no *lk (Table 6).

Non-inflecting -a- verbs with (as yet) no discernible middle-voice semantic: dalag to call
(someone).

This pattern discloses a more general theoretical issue. According to Comrie (1981: 161),
while the genuine derived causative [in Temiar, verbs in -r- ~ tr-] may be a productive process, the
derived anti-causative [in Temiar, verbs in -a-] will not be, since one cannot iteratively reduce the
degree of transitivity of a predicate: once it is intransitive, that is necessarily the end of the process.

But the Temiar middle voice in -a- does precisely that, as in the fully productive series gagl
(middle) gl (active) trgl (causative), where gl sit (Table 1) is indeed intransitive. Thus, the Temiar
middlecausative axis is not primarily syntactic, but semantic, referring to the contrast between inside
actions and external ones, with the neutral unmarked action, gl, represented by the root-form of the verb.
This suggests that in Temiar the syntactic valency schema is calqued upon the semantic relations, rather than
the other way round. So, while it may be syntactically irrational to further reduce an intransitive verb like
gl, it is still a meaningful thing to do, given the framework of understanding that Temiar-speakers operate
within.
21

The predominant orientational mode of Temiar culture is formed of the dialectical interplay of SELF
and OTHER. This dialectical pattern infuses their social interactional style (Benjamin 1994: 4447), their
religious life (Benjamin 1993a: 271273), and their musical structures (Roseman 1984).The closest that
Temiars ever come to talking explicitly about this aspect of personhood is in discussing the animating
subjectivities (souls) that are said to inhabit a wide range of beings, including people. It is then that their
ideas about the mutual entangling of SELF and OTHER become patent.
Such animate beings are said each to possess two souls, one associated with its upper part (the
head-hair roots of humans and animals, the leaves of trees, and the summits of mountains) and the other with
its lower part (the heart, breath and blood, the roots, and the subterranean mass). Dreamers and spirit-
mediums report that upper-body souls when encountered as spirit-guides (gonig) are like young men or
women in appearance, but that spirit-guides derived from lower-body souls appear as tigers. In other words,
upper-body souls are seen as familiar, domestic and SELF-like, while lower-body souls are seen as strange,
wild and OTHER-like. Yet it is a persons heart-cum-tiger soul, the hup, that is claimed to be the source of his
or her will and agency: it is ones hup that makes one do things or, alternatively, lacks the desire to do
something. Tigers, of course, are clearly OTHER. But it is also possible to perceive as other the usually
autonomous beating of ones heart (also hup)
22
or ones breathing (hmnum, the -n-infixed reduplicated form
of hup), since these can be directly monitored by the individual without their needing to be controlled. The
head-soul, rwaay, on the other hand, is clearly SELF-like in its association with the incessant but unobserved
growth of the hair the marker of bodily integrity. But the rwaay is also cast in the role of a patient-like,
non-controlling experiencer of whatever befalls the individual in dreams, trance and sickness one form of
which carries the (imperfective) verbal label rywaay, to suffer uncontrolled soul-loss.
Thus, for the Temiars, the controlling hup (the I) is an autonomous OTHER inside the person, while
the experiencing rwaay (the me) is an equally autonomous, but non-controlling, SELF. In other words, hup
beliefs imply that ones actions are at the same time something that one undergoes, while rwaay beliefs
imply that the things one undergoes (growth, dreaming etc) are at the same time ones actions (as in ones
dreams, which Temiars sometimes talk of as if they were activities that they can monitor). The individuals
empirical self or felt subjectivity is thus portrayed as a dialectical SELF-and-OTHER composite.
This cultural connection goes some way towards explaining why it is the objective marker -a- that
indicates the middle voice in Temiar, whereas many other languages express the middle-voice meanings
through a transparently subjective marking. The -r- in the Malay middle-voice prefix ber- also follows the

21
The increasingly common (British) English expression I was sat there is an interesting parallel to the
Temiar gagl sit.MID, in that it too is a (covertly) middle-voice construction involving an already intransitive verb,.
22
In stricter anatomical usage, as when butchering an animal, some Temiars associate hup with the liver rather
than with the heart.

same SUBJECTIVE pattern (Benjamin 2009: 306310). But since the subject of a middle-voice verb is as much
affected as affecting, the option exists for it to be encoded semantically as objective (-a-) rather than
subjective, even though that objectivity refers back to the grammatical subject.
This interpretation is supported by the manner in which the Temiar middle and causative voices
23

relate to each other as the two poles of the valency schema. (As already noted, there is no inflectional passive
voice in Temiar.) Whereas the middle voice carries a subjective meaning through an iconically expressed
incorporation of the syntactic object into the verb (as -a-), the causative voice expresses the objective
meaning of getting some other source or agency to do something through an iconically expressed
incorporation into the verb of a SELF-referring marker, the high consonant -r-. An example would be: rkaa
na-catk (door 3SG-close.MID) the door closed as opposed to ha-crtk [crtk] rkaa (2SG-close.CAUS.PFV
door) you closed the door (i.e. you caused the door to close). Thus, the dialectical SELFOTHER deixis
pervades the semantic and grammatical organization of the Temiar verb, and is given phonic expression
through iconicity.
24

I end this paper by showing that this pattern extends to Temiar nouns as well as verbs.
5. Middle-voice (unaccusative) nouns and nominalizations
The idea that middle-voice nouns can exist would seem to confound linguistic common sense.
However, if it is accepted that nouniness and verbiness are scalar variables rather than absolutes (Sasse 2001),
then the idea should be less surprising.
In Temiar there exist many disyllabic nouns with -a- in the first syllable, as well as deverbalized
forms combining -a- with the -n- nominalizer. On semantic grounds, a high proportion of these qualify as
middle-voice nouns, in that they refer to entities that can be thought of as simultaneously both acting and
acted upon, or as being both their own source and undergoer.
25
A categorized list of these nouns is given in
the Appendix. The largest category at least 135 items consists of mammals, reptiles, birds and arthropods.
As legged or winged animals, these make themselves move, and can therefore be thought of as their own
source and undergoer. The middle-voice morphology is therefore entirely appropriate. Fish names, on the
other hand (not listed in the Appendix), exhibit very few forms in -a-, perhaps because they are thought of
simply as undergoers of the flow of water in which they find themselves.
Fewer plant-linked words contain the vowel -a- in the first syllable. This somewhat reduces the
likelihood that they are interpreted as middle-voice nouns. But to the extent that these few words might be so
interpreted, I suggest that it would be on the grounds that plants, in shaping themselves, are the undergoers of
their own actions. Actions may seem inappropriate when talking of plants, but (as pointed out earlier)
Temiar animism does indeed accord agentive communicable-with subjectivity to several plant species.
Other categories of nouns with -a- in the first syllable include body parts, which can be thought of as
both moving and being moved.
26
A small but significant category consists of words for human relations,
which Temiars see as dialectical in character (cf. Benjamin 1994). The list of words for physical objects

23
Linguists do not usually refer to the causative as a voice of the verb, preferring to regard it as a derivational
formation. On intuitive grounds, however, it seems appropriate to include the causative in whatever grammatical category
the active, passive or middle are assigned to, and these have traditionally been labeled the voices of the verb. The purely
syntactic term valency would also be applicable, but for its failure to capture some of the semantic properties that are at
issue here. There is, however, a good partial precedent for the usage I am following here in Lehmanns reference
(1974:184), while discussing the middle and passive, to other meanings comparable to voices, such as the causative.
Jakobson (1957:4) too would seem to support this view: voice characterizes the relation between the narrated event and
its participants without reference to the speech event or to the speaker.
24
For further discussion of the dialectical mode of coherence that underlies Temiar culture, see Benjamin 1994:
4647.
25
Just as with verbs, there are many nouns with this shape that have no apparent middle-voice meaning at
least not that I can discern as of this writing. I list some of these at the end of the Appendix.
26
Diffloth (1974: 128131) discusses the same semantic issue in Semai, a close relative of Temiar, where
syntactic analysis shows that there too body-moves are not accorded a purely agentive interpretation.

in -a- contains many in which the middle-voice semantic is highly appropriate, in that they refer to entities
that hold themselves in place or which produce their own reaction (as explained in the Appendix below).
Examples of middle-voice nouns include:
layg night: because, as English puts it, night falls, happening spontaneously into existence of its
own accord. This word is both a noun (as in layg t last night) and a verb (as in hj na-layg
already 3SG-night, i.e. night has fallen).
sagub cloud: because clouds appear to bring themselves into being.
5.1 Etymological considerations
Etymological evidence suggests that -a- has been (deliberately?) inserted into certain Temiar words,
both nouns and verbs, at some stage in their history, to accord better with the middle-voice semantic
framework I have been discussing. Consider the following Mon-Khmer cognate-sets and derivational
series.
27

gat snail, cf. Shortos Proto Mon-Khmer reconstruction *gtoo, but on very restricted data
(Shorto 2006: 85): because a snail both supports and is supported by its own shell. Also, snails are
animals, and hence self-moving.
jalaa thorn, cf. Mon jrla, jla (Shorto 1971: 124), Jahai jla (Niclas Burenhult, p.c.) and
Semnam (Lanoh) jilaa (Burenhult & Wegener 2009: 291): because a thorn both snags and gets
snagged on peoples clothing.
jalk tree-top, cf. Old Mon clo /cl/ highest peak, apex, spire (Shorto 1971: 114): because a
tree grows its own summit.
kal stupid, silly, ignorant, dumb, cf. Middle Mon kamlau, kmlaw dumb (Shorto 1971:31):
because dumbness is something one undergoes despite oneself, not something one does.
kaii to court, flirt, cf. Old Mon (ka)i ~ gui to embellish (Shorto 1971:78): among Temiars
this is always reciprocal, the original meaning probably being to adorn each other.
28

kawaa kinsperson, cf. Spoken Mon kwa companion(s) (Shorto 1962: 227): in Temiar,
kinspersons are necessarily reciprocally so.
an chief, cf. Old Mon jnok to be great, high-ranking (Shorto 1971: 128): Temiar chiefs
become and remain such only through continuing dialectical relations with their followers.
sakl white-haired, grey-haired, cf. Old Mon sk, Modern Mon hk grey haired (Shorto 2006:
74): because hair becomes grey of its own accord.
5.2 Middle-voice deverbal nominalizations
Temiar verbs are regularly nominalized to form verbal nouns by the infixation or prefixation of (-)n-,
as displayed in Tables 14. Less productively, the infix -n- also occurs in a set of full nouns denoting
physical or abstract entities. An example is cnr knife, derived from the base-form of the verb cr to pare.
However, several such nouns relate more to the middle-voice of the verb, retaining or inserting the
characteristic -a- infix along with the middle-voice semantic that it marks, as illustrated in the following
examples:
bnatk, mnatk eyelet-loop (on a basket), from batk to make an eyelet-loop: because it both
supports and hangs from the strap that goes through it.

27
See also footnote 18, on the Mon-Khmer etymology of carh to go downhill.
28
A photograph of Temiar allo-adornment appears as plate 3 of Wavell et al. 1966, opposite p.19.

canaa food, from caa to eat: that which undergoes eating. Contrast this with the verbal noun
cnaa an eating.
canuu beater, from cuuk to beat, hammer: because a beater rebounds in the users hand.
Contrast this with cnr knife, above, which is not *canr because a knife acts more transitively
and consequently suffers no feedback.
29
(See also Matisoff 2003: 27.)
ganas tool for reaching fruit down from a tree, from gas to reach fruit down from a tree: because
a gas-ed fruit falls of its own accord, and the actor and tool undergo the fruit-fall. This is probably
an Austronesian loan: cf. Dempwolff 1938: 53, *gatentzweisein, Toba Batak, Javanese gas.
lan knowledge, from lk to know: because in a non-literate society knowledge is attained by
undergoing experience, not through active study.
mnanuu size, bigness, from mnuu to be big: that which has become or been made big.
panh shamanic dancing space, from ph to hold a sance: the portion of the house where both
the shaman and the house itself undergo possession by a spirit-guide, which is also sometimes
referred to as ones ph.
snalg marriage, married state, undergoing marriage, from slg to marry. Contrast this with
the verbal noun snlg marrying.
References
Arce-Arenales, Manuel, Melissa Axelrod & Barbara A. Fox. 1994. Active voice and middle diathesis. In:
Barbara Fox & Paul J. Hopper (eds), Voice: Form and Function, Amsterdam: Benjamins, pp. 121.
Baldi, Philip. 1974. Reciprocal verbs and symmetric predicates. In: Michael W. La Galy et al. (eds), Papers
from the Tenth Regional Meeting, Chicago: Chicago Linguistic Society, pp. 1726.
Benjamin, Geoffrey. 1976a. Austroasiatic subgroupings and prehistory in the Malay Peninsula. In: Philip N.
Jenner, Laurence C. Thompson & Stanley Starosta (eds), Austroasiatic Studies, Part I, Honolulu:
University Press of Hawaii, pp. 37128.
. 1976b. An outline of Temiar grammar. In: Philip N. Jenner, Laurence C. Thompson & Stanley
Starosta (eds), Austroasiatic Studies, Part I, Honolulu: University Press of Hawaii, pp. 129187.
. 1993a. Temiar. In: Paul Hockings (ed.), Encyclopedia of World Cultures, Volume 5: East and
Southeast Asia, Boston: G. K. Hall/MacMillan, pp. 265273. (Also: www.everyculture.com/East-
Southeast-Asia/Temiar.html.)
. 1993b. Grammar and polity: the cultural and political background to Standard Malay. In: W. A.
Foley (ed.), The Role of Theory in Language Description, Berlin: Mouton de Gruyter, pp. 341392.
. 1994. Danger and dialectic in Temiar childhood. In: Josiane Massard-Vincent & Jeannine Koubi
(eds), Enfants et Socits dAsie du Sud-est, Paris: LHarmattan, pp. 3762.
. 2001. Orang Asli languages: from heritage to death? In: Razha Rashid & Wazir Jahan Karim (eds),
Minority Cultures of Peninsular Malaysia: Survivals of Indigenous Heritage. Penang: Malaysian
Academy of Social Sciences (AKASS), pp. 101122.
. 2004. Aslian languages, Aslian: characteristics and usage. In: Asmah Haji Omar (volume editor),
The Encyclopedia of Malaysia, Volume 12: Languages and Literatures, Kuala Lumpur: Archipelago
Press, pp. 4649.
. 2009. Affixes, Austronesian and iconicity in Malay. Bijdragen tot de taal-, land- en volkenkunde
165: 291323 (also www.kitlv-journals.nl/index.php/btlv/article/view/3653/4420).
. 2011. Temiar morphology (and related features): a view from the field. 49pp. Unpublished
typescript.
Benveniste, Emile. 1971. Problems in General Linguistics. Translated by Mary Elizabeth Meek. Coral
Gables: University of Miami Press.

29
The morphophonological changes involved in the derivation of canuu from cuuk and of lan from lk are
explained in Benjamin 1976b: 143.

Blagden, C. O. 1906. Language. In: W. W. Skeat & C. O. Blagden, Pagan Races of the Malay Peninsula,
volume 2, London: MacMillan, pp. 379775.
Broselow, Ellen & John J. McCarthy. 1983. A theory of internal reduplication. The linguistic review 3: 25
88.
Burenhult, Niclas & Claudia Wegener 2009. Preliminary notes on the phonology, orthography and
vocabulary of Semnam. JSEALS: Journal of the Southeast Asian Linguistics Society 1: 283312.
Carstairs-McCarthy, Andrew. 1992. Current Morphology. London/New York: Routledge.
Comrie, Bernard. 1981. Language Universals and Language Typology. Oxford: Blackwell.
Diffloth, Grard. 1972. Notes on expressive meaning. In: Paul Peranteau et al. (eds), Papers from the 8
th

Regional meeting, Chicago Linguistic Society: 440447.
. 1974. Body moves in Semai and in French. In: Michael W. La Galy et al. (eds), Papers from the
10
th
Regional Meeting, Chicago Linguistic Society, pp. 128138.
. 1975. Les langues mon-khmer de Malaisie: classification historique et innovations. Asie du sud-est
et monde insulinde 6 (4): 119.
. 1976a. Minor-syllable vocalism in Senoic languages. In: Philip N. Jenner, Laurence C. Thompson
& Stanley Starosta (eds), Austroasiatic Studies, Part I, Honolulu: University Press of Hawaii, pp.
229247.
. 1976b. Expressives in Semai. In: Philip N. Jenner, Laurence C. Thompson & Stanley Starosta
(eds), Austroasiatic Studies, Part I, Honolulu: University Press of Hawaii, pp. 249264.
Frampton, John. 2004. Distributed Reduplication. Unpublished monograph, February 2004. Northeastern
University. Downloaded from: www.math.neu.edu/ling/DR/DR_March1_04.pdf.
Gafos, Adamantios I. 1998a. Eliminating long-distance consonantal spreading. Natural language and
linguistic theory 16: 223278.
. 1998b. A-templatic reduplication. Linguistic enquiry 29: 515527.
. 1999. The Articulatory Basis of Locality in Phonology. New York: Garland.
Hendricks, Sean. 2001. Reduplication without prosodic templates: a case from Semai. Journal of East
Asian linguistics 10: 287306.
Idsardi, William J. & Eric Raimy. N.d (2006?). Representational economy. No pagination [73 pages].
Downloaded from: http://mendota.english.wisc.edu/~raimy/papers/idsardiraimyrepecon.pdf.
It, Jungko. 1989. A prosodic theory of epenthesis. Natural language and linguistic theory 7: 217259.
Jakobson, Roman. 1957 [1971]. Shifters, verbal categories and the Russian verb. Pages 130147 in his
Selected Writings, Volume 2. The Hague: Mouton.
& Linda R. Waugh. 1979. The sound shape of language. Brighton: Harvester Press.
Lehmann, Winfred. 1974. Proto-Indo-European Syntax. Austin: Texas University Press.
Levin, Beth & Malka Rappaport Hovav. 1995. Unaccusativity. Cambridge MA: MIT Press.
Matisoff, James A. 2003. Aslian: Mon-Khmer of the Malay Peninsula, Mon-Khmer studies 33:158.
McCarthy, John. 1982. Prosodic templates, morphemic templates, and morphemic tiers. In: Harry van der
Hulst & Norval Smith (eds), The Structure of Phonological Representations (Part 1), Dordrecht:
Foris, pp. 191223.
Means, Nathalie. 1998. TemiarEnglish, EnglishTemiar Dictionary. (Edited by Gordon P. Means.) St Paul:
Hamline University Press.
Pinnow, Heinz-Jrgen. 1959. Versuch Einer Historischen Lautlehre der Kharia-Sprache. Wiesbaden:
Harrassowitz.
Raimy, Eric. 2000. The Morphology and Morphophonology of Reduplication. Berlin: Mouton de Gruyter.
Roseman, Marina. 1984. The social structuring of sound: the Temiar of Peninsular Malaysia.
Ethnomusicology 28: 411445.
Sasse, Hans-Jrgen. 2001. Scales between nouniness and verbiness. In: Martin Haspelmath et al. (eds),
Language Typology and Language Universals, Part 1. Berlin: Walter de Gruyter, pp. 495509.
Shaw, Patricia A. 1994. The prosodic constituency of minor syllables. In: Proceedings of the Twelfth West
Coast Conference on Formal Linguistics: 117132. Stanford: Stanford Linguistics Association.
Shorto, H. L. 1962. A Dictionary of Modern Spoken Mon. London: Oxford University Press.
. 1971. A Dictionary of the Mon Inscriptions from the Sixth to the Sixteenth Centuries. London:
Oxford University Press.
. 2006. A Mon-Khmer Comparative Dictionary. Canberra: Pacific Linguistics.

Sloan, Kelly. 1988. Bare-consonant reduplication: implications for a prosodic theory of reduplication.
Proceedings of the Eleventh West Coast Conference on Formal Linguistics, 319330. Stanford:
CSLI Publications.
Spaelti, Philip. 1997. Dimensions of variation in multi-pattern reduplication. Unpublished PhD thesis in
Linguistics, University of California, Santa Cruz.
Spencer, Andrew. 1991. Morphological Theory. Oxford: Blackwell.
Ter Mors, Christine. 1983. Affix to X. Broselow & McCarthy versus Marantz: a reinterpretation. The
linguistic review 3: 275297.
Walther, Markus. 2000. Temiar reduplication in one-level prosodic morphology. Proceedings of the
SIGPHON 2000 Workshop on Finite-State Phonology: 1321, Luxembourg, 6 August 2000.
Downloaded from: www.arxiv.org/pdf/cs.CL/0008015.
Watkins, C. 1976. Towards Proto-Indo-European syntax: problems and pseudoproblems. In: Sanford B.
Steever et al. (eds), Papers from the Parasession on Diachronic Syntax, Chicago Linguistic Society,
pp. 305326.
Wavell, Stuart et al. 1966. Trances. London: Allen & Unwin.
Yap Ngee Thai. 2009. Non-exhaustive syllabification in Temiar. JSEALS: Journal of the Southeast Asian
Linguistics Society 1: 267282.


Appendix
Lists of deponent verbs and middle-voice nouns
This list was re-checked in the field in October 2010, but it is still not exhaustive. Malay loanwords have
been excluded. Many of the words listed here were characterized by my respondents as old (manah). Other
words that I had collected previously sometimes just two years earlier! were now judged not to exist at all,
and are accordingly not listed here. These changing evaluations may indicate that under modern conditions
such middle-voice meanings are thought to be less noteworthy than previously.
Verbs
adees to exercise (ones body), keep moving around: one does this to oneself
aluur to shoot/be shot backwards, of felled branch: it undergoes its own action
aro to talk nonsense, to joke; to flirt, entice: a socially reciprocal activity
ayir (1) to plait the interwoven bottom on a bag: the strands hold themselves in place
babh to act undecidedly: one undergoes ones own indecision
bacuuh to imitate (an animals sound); to echo: because the animal sounds through the imitator; because the
sound comes back upon the sounder
bagd to try something new for the first time: both an experience and an activity
bak set a bak trap: such traps are self-springing
ball to roll something up in a wrapper: because the wrapping then holds itself in place
balaaw to fail to catch game when hunting: the hunter undergoes the failure
balec to abuse, speak angrily to someone: such behavior rebounds on the abuser
balc to not believe (religion); to refuse (to do something): such behavior rebounds on the refuser
batk to make a loop (bnatk) (for the straps of a basket): the loop holds itself in place (see the main text)
bawk to bend a springy object to shape: the bender both pulls and is pulled by the object
cabl to join things together by insertion: the parts both join and are joined
cacii to dance (standing still while moving arms and hands): the dancer moves her body while
experiencing her centre of gravity unmoved
cadog to decorate ones hair or earlobe with flowers or ornaments: the inserted decoration holds itself in
place
cad to sit cross-legged: the legs hold each other in place
cakd to be sticky, to stick: sticky objects both stick and are stuck
cal
b to have specks in ones eye: an undergone experience

cal to pour away: the pourer senses the vessel getting lighter, thereby experiencing the pouring
calh to fruit, set seed (rice, millet): the plant undergoes the fruiting
cana to beat a gong: the beating rebounds through the arm and ear; once started, the gong keeps itself
sounding
carh to descend a slope: one undergoes ones own descending (see the main text)
card to be interlocked: the parts both lock and are locked
catd to be physically obstructed (by a crowd or object): one both obstructs and is obstructed
cayaa to open ones legs wide (as in childbirth or intercourse): the legs are held in place by their own
friction against the flooring or ground
gabag sing: the singer both sings and is sung through (see the main text)
gacuuh to collide (a body part) into something: the body both knocks and is knocked
gadis to hurry around requesting things: one undergoes the momentum of ones own movement
galag to eat greedily; to act enthusiastically: one is overcome by ones own greed or enthusiasm
galk to scratch, tickle: the nails are resisted by the flesh
galook to beat bamboo stampers (in song sessions): the bamboos bounce back into the hands
gapid to be trapped or tightly squeezed (as in a tree-fork): one undergoes the experience (semi-deponent:
grpid to squeeze, but no *gpid)

gatah to chip away (with a small knife): the action rebounds on the hand
hahj to burst out laughing: overcome by ones own uncontrolled laughter
halab to transport/be transported downriver: the raft is transported by the rivers flow (see the main text)
halg to laugh (in company): because such laughing is catching
harj to cease raining: the rain appears to stop itself
haya to nearly do something, narrowly avoid doing something: experienced rather than carried out
deliberately
jajeeh to be abundant, in excess; to leave a surplus: one finds oneself unable to finish (or reproducing too
fast) through no positive action of ones own
jalb to set (sun): the sun appears to do so of its own accord
jawg to stride: one gets caught up in the rhythm of the movement
kabd to hug tightly (climbing a tree or riding piggy-back): climbers hold themselves in place
kabd to send off tightly hugging side-shoots (plant, as on a strangling fig): the plant forms itself
kaceed to not know: ignorance is a state one finds oneself in, not through any actions of ones own
kac
d to boil over, flow over (water): the vessel undergoes the overflow; to give fire to (from flame pr
spark); the object undergoes the fire
kac
j to crush (a louse etc) with ones fingernail: one experiences the crunch of the insect
kadeeb to climb something, clasping with hands and feet: the climber both holds on and is held
kakh to beat with hand or stick, without cutting: the action rebounds on the beaters hand
kalag to eat greedily: the eater is driven by his own greed
kaleek to prop: it both supports and is supported (see the main text)
kaln to feel pity: the pitier both extends pity and experiences it
kalt to feel sorrow, mournful: the sorrower both extends sorrow and experiences it
kamum chew: the food resists the chewing.
kab ~ kalb to blink: both an action and something experienced
kaii to court, flirt: a reciprocal action (see the main text)
kapej ~ kapeej to claw (with both paws, like a tiger or bear): the claws undergo resistance
kapk to kill by beating: the action rebounds on the killers hand
kapus to overflow: the vessel undergoes the overflow
karaaw to cry out: an emotional response that happens to the crier
karh to dry off, dry out: both an action and a process
lalee to assimilate oneself; ethnically mixed (person, population); Gb lalee person with a Temiar mother
and a Malay father; Kuy lalee type of Malay language used in some song-ritual lyrics: assimilation is
something one both does and undergoes
lapj to be encrusted (eye after sleep): it happens to the waker
lapood to be soaking wet: this is undergone
malar to follow around listlessly (like a child): the child cannot help itself
mam to suffer a kind of headache (believed to be caused by dew); an undergone experience; also a
middle-voice noun
aaah to feel sympathy or sorry for someone: one cannot help doing so
pacoh to hurry on ahead, hurry up: one gets caught up in the rhythm of the movement
pasuu to not like someone: regarded as rebounding onto the initiator
payaa to not want to court someone, reject someone: regarded as rebounding onto the initiator
ragh to travel in a hurry: one gets caught up in the rhythm of the movement
sab to catch (of a trap): the trap releases itself
sagoob to hold food in a bulging gullet (snake, lizard): the gullet and food hold each other in place
saloh to lodge, stay with someone: regarded as a reciprocal activity
saluur to faint; to die: an undergone process
sapood to wrap: because the package keeps itself wrapped
sarg to give way (soft ground): the ground and the walker undergo the subsidence
sath to collapse, break off, release (of fall trap): the object undergoes its own collapse (see the main text)
tabag to hang something up for storage: the item hangs itself
tad to hold ones hand out: an action in expectation of receiving something; or an action in muscular
dynamic equilibrium

tajeeh to suspect (whether): suspicions come uninvited
tajl to hang (trans): the item hangs itself
takil to throw something strongly; to bowl (as in baseball, cricket): the thrower experiences the follow-
through momentum
tak to cut bamboo (not wood) into short sections: the cutters experiences the rebound
takuu to hurl (a spear); to travel quickly (airplane): it covers both active and middle meanings to hurl and
be hurled
talay to postpone (an activity): one is then under obligation to do it later
talr (foot) to slip or slide on muddy ground: the foot undergoes the slipping
taloot to be untrustworthy, deceitful: this will rebound on the liar
tal to try food for taste, try out a blowpipe, test something: one experiences the sensation of newness
tara to stop: one brings oneself to a stop
tarah to plane flat: the hand undergoes the resistance of the wood
tarl to fall off (dried up umbilical cord): it undergoes its own action
tays to point: because the pointer experiences the arms weight
tayd to lift (a heavy object): because the lifter both lifts and undergoes the pull of the weight
yayah to be crazy, wild: one undergoes the emotion
Stative verbs, Adjectives
These are conditions or circumstances that the source appears to undergo.
alooy unripe, fresh
babuh (1) rotten (wood)
bahul (1) to be a big eater: he is driven by his own greed
ga~ ga to be shrunken with hunger (belly)
gadal ~ kadal hard pan (soil), firm (ground), caked (powder) (semi-deponent, ~ kldal, but no *kdal)
gahl out of breath, weary
galah straight (hair)
gatw thin (body, cane),
gawk skinny
gayuh rough or soft (stony or overgrown ground)
haraaw (1) hairy, frizzled (hair, fur, spines, clothing)
hay
light (weight)
jalg restless, promiscuous (staying in a different place each night)
jamii swollen-cheeked
kal stupid, silly, ignorant, dumb (see the main text)
kark to feel very cold; rigor mortis
karuk withered, dead (plant), weak, near death (person)
katd to swell (a boil, a tree-canker)
lapood soaking wet
latah bald in front
latb bald on top (with hair at the sides)
manah to be old, former (of things, not animals or humans)
papah to be the wrong way round, upside down
pap crazy, mad, insane
rahm round-faced
ratih diligent (in work): carried away by ones enthusiasm.
rawj finished, used up
sagl thirsty
sakl white-haired, grey-haired (see the main text)
saraa ma- to depend on someone, to be up to someone to decide: dependency is mutual
sayeep dead, dried out (bamboo)
takl giggly, laughing a lot: giggling is uncontrollable

talr to be slippery (ground)
wawar to be out of true alignment
Nouns
Mammals and reptiles
These are all legged creatures that move themselves through their own actions.
It has proved difficult to identify many of these animals more closely. Many of the animal names are
avoidance names or taboo names, and are therefore not necessarily the common ones.
abir a squirrel
ab a land rat
acaam a large squirrel
aceel a large squirrel
acl a large squirrel
acoh common grey-bellied squirrel
ad a small squirrel
agj ~ agc common Malay squirrel, plantain squirrel
ahl flat-headed cat
ajh rusa deer (Cervus timorensis)
aj
r, ajoor common tree-shrew

akuub forest squirrel
alaam elephant
alaay ~ alaaj elephant
alp elephant
alul elephant
amaah a small frog
am goat-antelope
amug a white squirrel
aaa a small red-tailed squirrel
app a large civet cat
ap pig-tailed macaque (Macaca nemestrina)
ar barking deer (Muntiacus muntjak)
ataah horse-tailed squirrel
ata ~ ataa tiger
atd elephant
badt bearded pig
bagt monitor lizard (Varanus rudicollis)
bakaa small-toothed palm-civet, bear civet
bapaak a white-bellied dark-backed squirrel
barw tapir
bash white-eyed gibbon
batw a gibbon
bawaaj pig-tailed macaque
ca
a small house-mouse
cad a small squirrel
capg solitary large male (of monkeys)
dalk a lizard
dari soft-shelled river turtle
hadaa short-tailed mongoose
hagaab two-horned rhinoceros
har monitor lizard (Varanus salvator)

hayum bamboo rat
jaj common palm civet; tiger-civet
ja Mullers rat
kab a frog sp.
kabuc monitor lizard (Varanus salvator)
kabuk a green snake
kaj flat-backed land tortoise
kahuul tadpole
karj a river-turtle, small box-tortoise
kasi rusa deer (Cervus timorensis)
kaweeb Malayan sun bear
kayii flying lemur
mamuug tiger
pari a monitor lizard
payd a monitor lizard
sakl ~ sagl cream-colored giant squirrel
ta wild pig
tabg bullfrog
tabk ~ tabk dusky leaf-monkey
tagaat a snake
tahr (1) striped ground squirrel
tajuu snake
talam elephant
tapl pygmy flying squirrel
tarook a lizard
tawh white-handed gibbon
Birds and bats
These are all winged creatures that move themselves through their own actions.
ahah great slatey woodpecker
ak
b stonechat
akg red-heaed trogon
akuul trogon (generic)
at
b grey-chinned minivet
bajw cotton teal
baryt (a kind of bird?)
bayj a small night-calling bird
caba black-crested yellow bulbul
cabw chestnut-collared kingfisher
cacr red-headed tailor bird
daw lesser short-wing
gagoob large scimitar babbler
kaka kingfisher
kakeh hornbill (various kinds)
kakuuh black hornbill, white-billed hornbill
kasaa grey-headed tree-babbler
kasar black-headed bulbul
kaw
d fruit bat, flying fox

lasar large bat, flying fox
pah black wood-partridge
payh ferruginous babbler
rabl weaver finch

sabaat hawk owl
sagur a small bird
sayah (1) red-headed tree-babbler
sayl mountain nun-thrush
sayl false vampire bat
taj hornbill sp., wrinkled hornbill
tadoor red jungle fowl
tagk gold-whiskered barbet
tahr southern pied hornbill
tahr grey and buff woodpecker
tald Scops owl
tapr white-eared fruit bat
tatoh maroon woodpecker, greater yellow-naped woodpecker
tayt grey wagtail (or blue-throated bee-eater?)
Arthropods and snails
These are all creatures that move themselves through their own actions.
bahul (2) a small dung beetle
bayj a bird
capood a very large black, or blue or striped fly; a large biting fly
capg polydesmoid millipede
cawaas ~ caws earwig
galul a large black mosquito
gard a sunset-chirping cricket
gark a night-chirping cricket
garuuc termite
gas red stinging ant
gat mite
gat snail(shell) (see the main text)
jaleed firefly
jarb a small noisy outdoor cricket
jawiis a seasonal cicada
kabd ant
kajk a black biting ant
kasd fire-ant
lawh a large red swarming tree-ant
maaay scorpion
padaaw hornet
saley jewel beetle
tabl black honey bee
tala a large beetle
taly black millipede
talk bumble bee, carpenter bee
tan dragonfly
taro house lizard
tawg butterfly
tawiik spider
wawah a moth
yayeed an evening-sounding cicada

Body parts
Body parts can be thought of as both moving and being moved.
ayir (2) itchy scalp, from louse feces
bakoh male genitals (sometimes penis, testicles)
balk animals beak or horn
cawk head, skull
jaka ~ caka lower jaw (also: overhang of a roof)
kabooj large vesicular swelling
kadg hollow of the knee-joint
kalar throat, glottis, esophagus, adams apple
kalej testes
kalr snout of pig, bear, cat, etc.
ka elbow
kap
cheek
karb sternum
karl knee
kato knee-cap
kaway wing
kay little finger
kayood fetus: an Other within ones Self
lag caul placenta, afterbirth; also new-born baby: an Other within ones Self (Benjamin 1994: 52)
pac
still-born fetus
panik navel
sabook windpipe
sakb corpse
sapal upper arm
taboo thumb, big toe
tan neck
tapaag palm, sole; hand of bananas
tapaar back of the hand, instep of the foot
Plants and plant parts
Plants, in shaping themselves, can be regarded as the undergoers of their own actions.
adg a tasty wild tuber
ajl a cane
apoos wild ginger
asaad a large squash
ash millet (taboo name)
awaat bamboo
awn bamboo
babuh (2) a small toadstool
badk jelutong tree
bajaaw a fruiting vine
bayas a palm with edible pith
bayoor a large secondary-forest tree
cadag a long-leafed plant used for plaiting ritual crowns
cakoob bark (of tree)
cah pulasan fruit
galook large rotten limb of a tree, about to fall
gareed a tree (provides the barkcloth used by menstruating women)

gayk a wild fruit
haog a fruit-bearing forest tree
hakr a wild seasonal fruit
hayul generic term for staple food crops (non-leafy)
haraaw (2) a forest flower
kabup calyx, bud
kacuuh a bitter leaf eaten with betel
kalok ~ kalk the bark eaten with betel chew
karuu tree-resin illuminant: it exudes itself; it keeps itself alight.
katak node of bamboo; vertebra: it separates the other parts, but is held in place by them
kawk a medicinal herb
labk rolled-up leaf bud, young shoot (of wild ginger, bamboo)
lal a tree
manaar a cane
pasg a species of cane
pal firewood
raniik a wild seasonal fruit
rarh a wild seasonal fruit
sakb the spiny sheath on bamboo
sakool a long leaved vegetable
sa a bush, the stem of which is pounded for medicine
sag a mountain-growing bush
sarg a wild yam
sayah (2) the chaff after winnowing rice grains
taba springy branch
tamud a sweet-smelling herb
tang a leafy vegetable with a spicy taste
Human and spiritual relations
Temiar social relations are thought of as dialectically generating each other.
ayad gang, group (of friends)
balu widow, widower
cac grandchild, younger siblings child
Karey thunder, Thunder deity: both cause and result
kawaa kinsperson (primarily consanguineal) (see the main text)
lay ghost; also used of terrorist during the Malayan Emergency
an chief (see the main text)
lag newborn baby (also used to label various perinatal misfortunes)
pacg a spirit-invasion disease, causing a prickly sensation: both cause and result.
paley spirit-mediums hut (strictly, the palm-leaf used in its construction): it undergoes possession by the
tiger-spirit
papd infant(plural pdpd ~ pnpd, but no singular *ppd)
sabat a convulsive attack (associated with childhood and childbearing) (see Benjamin 1994: 5455)
sap pollution caused by recent death
sar ghost of deceased
say
d young child (plural sdy
d ~ sny
d, but no singular *sy
d)
tajr watery fingertip manifestation of ones spirit-guides presence
tanig a gout-like disease (caused by interaction with a river-spirit): both cause and result

Physical objects and processes
These are thought of as being simultaneously their own source and undergoer.
abat sarong: it holds itself in place
alg plaited storage bag: it forms its own shape as it fills
ap back-basket: it is self-hanging
apil mat: it lays itself down
apk tobacco pouch: it takes its own shape
bak large noose trap, set on tree branch: it springs itself
baliik sky, position above: it appears to be self-supporting
baraa main beam (of house): because it holds itself in place
bawur fishing rod: because it both pulls and is pulled
cacuh woven roof thatch: it holds itself together
cagl pond: it is self-forming
gada winnowing tray: it throws the grains and is thrown by them when they fall
gagid middle: it falls between two limits while holding them apart
gahool a depression in the ground, valley: it is self-forming
galeed open-weave rotan basket (for carrying water-tubes): it is self-hanging and weighed down by its
contents
gal
l glowing ember: it keeps glowing of its own accord

gask brochette, sliver of wood for cooking on, splinter of bamboo (for tattooing): it both penetrates and is
surrounded by what it penetrates
jalaa thorn: it both snags and gets snagged on peoples clothing (see the main text)
jalk tree-top: a tree grows its own top (see the main text)
lamuu bent-over sapling-spring in various kinds of trap: it is held in dynamic equilibrium
layg night (both a noun and a verb): night falls, happening spontaneously into existence of its own accord
(see the main text)
raboo a large backbasket: it holds itself in place
ran back-basket: it holds itself in place.
sagub cloud: clouds appear to bring themselves into being (see the main text)
salg a kind of backbasket (for use when trapping): it holds itself in place
samg blowpipe wadding: it both blows and is blown
sap section of a house affected by death-pollution: it both pollutes and is polluted
saroog plaited rice-bin: it takes its own shape when filled
takoo pool, pond: it is self-forming
tak small container (for bait, wadding): it is self-hanging
Unclassified: not obviously deponent or middle-voice
A sampling of residual verbs and nouns in -a- that cannot as yet be explained in the terms discussed in this
paper
abaag split-bamboo internode used as eating dish
acag to plan to do something
alh to guess
ams small
apt short
arap a possession
dalag to call someone
halh to guess
kamaay ~ kamy to store food (against future hunger)
kan we excl.
las dirty

lagoh dark leafy vegetable, like a bayas leaf
lakb to fold (trouser cuff, corner of a page)
lalah open terrain
law flower-odor (ritual language)
lawg to mix and scoop food up with the fingers
lawud to cook a stew
mark fish-weir platform
pal log of firewood
pani eventually, in future
rages to serve oneself bit by bit from the different dishes at a communal meal
sarag to transplant (tobacco, wet rice)
tabh to prepare something (for use)
tahr (2) hole (animals, in a nest or the ground)
takaah slope
targ spear
tayd to pick something up between fingers and thumb

Ferlus, Michel. 2011. Toward Proto Pearic: Problems and Historical Implications. In Sophana Srichampa and Paul
Sidwell (eds.) Austroasiatic Studies: papers from ICAAL4. Mon-Khmer Studies Journal Special Issue No. 2. Dallas, SIL
International; Salaya, Mahidol University; Canberra, Pacific Linguistics. pp.38-51.
38
Toward Proto Pearic:
Problems and Historical Implications
Michel Ferlus
Independant Researcher (ret. CNRS, France)
Abstract
The reconstruction of Proto-Pearic phonology allows us to consider the following
hypothesis: The contrast /tense vs lax/ in Vietic, Katuic and Pearic was formed under the
influence of Ancient Chinese along the trade route leading from North-Vietnam to the gulf
of Thailand.
1. Introduction
This text will examine of the essential facts of Pearic: the problem of the two-way shifts of the
voiceless initial plosives, the origin of the final glottal stop, and the origin of the two layers of registers
(creaky and breathy). I will present a chronology of phonetic changes and an attempt an explanation of their
propagation along ancient trade roads through Southeast Asia.
30

2. Ethnonymes in Pearic subgroup
Chong, also Song/Xong/Kasong, originate from *k as the genuine autonym of Pearic
populations. It is attested as Tchouang within Tchouang thieves (zhung zi ) in the Tcheou Ta-kouan
description of Cambodia, from the end of 13rd century (Pelliot 1902: 156; 1951: 70). In modern Khmer: ja
c 1. barbare, 2. nom de tribu demi sauvage (Guesdon 1930); and Chong (name of a tribe),
barbaric, wild (Jacob 1974) with a strong derogatory connotation. The meaning barbaric perhaps allows a
connection with zhung Zhuang people, also zhung/chung .
Por and Pear originate from Skt. vara- color, caste acording to two treatments in Khmer.
Por: from Skt. vara- through the treatment *br> *br> *br. It is attested in Khmer
dictionaries as color, appearance; bar pr> p (Guesdon 1930: 1203).
Pear: from Skt. vara- through the literate treatment *bar> khmer br / ba(r) par >
p (Ferlus, 1981). Its use by Khmer and French administration justifies the present name of the so-called
Pearic subgroup.
Pol: from Skt. bala army, guard. The Pol were at the disposal of the King for the guard of
monuments and other places (Brengues 1905), they were composed of war prisoners, convicts and mountain
peoples. Contrary to appearences, Pol is not an alternative of Por.
Samr, also Somre/Somr/Somray smre/smraj: formed of sre field, cultivate infixed by
-m-; meaning cultivators.

30
Abbreviations used in this paper: MK: Mon-Khmer; PMK: Proto Mon-Khmer; PP: Proto Pearic; EPP: Early
Proto Pearic; LPP: Late Proto Pearic; OC: Old Chinese; MC: Middle Chinese. T: tense; L: lax.
Michel Ferlus 39

Saoch, or the Chung of Kompong Som. This exonym is used only by Khmer speakers. In Chung, it
would be *soc. Saoch has been connected to the Khmer s'c
soc scarlet fever (Diffloth, see in Isara

2007: 60). Guesdon (1930: 1841) gives pimply. Derogatory ethnic names are rare, so why would the
Khmer, who used Pear to name other groups, have marginalized Chung people with the derogative Saoch?
Saoch was also thaicized into Kha Saut and Ut ut in King Rama Vs travelogue to name the Chung Yul
of Thailand (Isara 2007: 60, 26-27).
Khamen Boran of Pursat (Bastian 1868: 264-6) spoke a Pearic language although this designation
usually names the Khmer Daeum (Martin, personal communication).
The exonym Suay of Kompong Speu Pear is shared to name some Katuic population.
3. Pearic Languages and dialects:
There are four generations of linguistic data:
Before 1900: short vocabularies.
From 1900 to 1970: consistent vocabularies collected by attentive investigators but not linguists,
French in the majority, knowing the Khmer language or working with Khmer assistants. These vocabularies
give a satisfactory idea of the consonantism and, a little less, vocalism.
The data of Marie Martin, which recognizes the existence of register features.
The 1980s marks a transition with the arrival of professional linguists in Pearic studies. We have
from now good data and scientific analyses. The register system is now clearly described (e.g. Huffman 1985;
Theraphan 1984, 1991).
Data of the pre-linguistics period make it possible to date the shifts of finals *-r, *-l and *-s. French
investigators normally distinguish clearly *-r and *-l which are noted -l and -rr by Baradat. The final spelled
-s is not consistent, it represents *-s as well as *-h. It can be explained by Khmer writing in which final -s is
generally pronounced -h these days. The rules of Khmer spelling were translitterated into Latin spelling, and
there is also, in certain authors, an improper use of the empty -r, as in khmer akar k husked
rice in which -r is not significant.
The Proto Pearic of Headley (1985) was elaborated with pre-linguistic data. Despite this handicap,
this work remains a milestone in the Pearic studies and a good basis for further researches.
Pearic languages Authors Date
Chong Crawfurd 1828
Samreh dAngkor Bastian 1868
Khamen boran Bastian 1868
Xong Bastian 1868
Xong Garnier 1873
Khamen boran Garnier 1873
Kouys - Porrh Harmand 1878-79
Saoch Pavie 1881-82
Smr (Siem Rap), Pors (Pursat) Moura 1883
Sauch [Leclre] 1900
Porr (Pursat) Brengues 1905
Saoch Pannetier ?
40 Towards Proto Pearic

Per de Kranhung Pannetier ?
Souy Pannetier ?
Saoch Mntrier 1926
Chawng (Ban Dan Champon, Tratt) Isarangura 1935
Pear des Cardamomes (Phnom Krevanh) Morizon 1936
Chng de Siam (=Chong of Trat) Baradat 1941
Pr de lOuest Baradat 1941
Pr de lEst (Peam Prus) Baradat 1941
Kompong Suoi de Kompong Speu Baradat 1941
Pr de Thom Baradat 1941
Saoch (Kompong Som) Ellul 1968-
Saoc (Komong Som) Purtle (in Headley) 1970-
Chong l (Chanthaburi) Martin 1974
Chong hp (Chanthaburi) Martin 1974
Smray (Kranhung, Batdambang) Martin 1974
Smree (Peam Prus, Pursat) Martin 1974
Chong (Klong Phlu, Chanthaburi) Surekha 1982
Samree (Borai, Trat) Theraphan 1984
Chong (Klong Phlu, Chanthaburi) Huffman 1985
Chong (Klong Phlu, Chanthaburi) Sirikarn 1987
Chong (Klong Phlu, Chanthaburi) Saifon 1991
Chong (Klong Phlu, Chanthaburi) Theraphan 1991
Chong (Klong Phlu, Chanthaburi) Edmonson 1996
Kasong (Khlong Saeng, Trat) Kunwadee 1996
Chong (Klong Phlu, Chanthaburi) Siripen 2001
Samre (Trat) Pornsawan 2001
Chong (Klong Phlu, Chanthaburi) Isara 2002
Kasong (Khlong Saeng, Trat) Noppawan 2003
Kasong (Khlong Saeng, Trat) Sunee 2003
Chong (Klong Phlu, Chanthaburi) Isara 2004
Chung Yul (Kampong Som) Isara 2007
Chung Yuy (Thung Na, Kanchanaburi) Isara 2007
Chung (Kanchanaburi & Kampong Som) Isara 2009
Table 1: Chronology of Pearic data
4. Register System of Pearic
Acoustic analyses of Chong by Theraphan (1991) once and for all demonstrated the existence of a
four registers system in Pearic that preceding works foresaw to some extent (Martin 1974a; Surekha 1982;
Huffman 1985).
Michel Ferlus 41

The system combines the modal feature /clear/ with the marked features /breathy/ and /creaky/ to
form four syllabic combinations.
R1 [CVC]
Clear modal
R2 [CVC]
clear-creaky

R3 [CV C]
breathy
R4 [CV C]
breathy-creaky
Table 2: Pearic registers
Examples (Siripen 2001):
R1 kak armpit, tak span, ck pound, kla branch
R2 ktak tongue, knak crow, ks ladder, kla owl
R3 ta k water, kla ear, c k shine, kp k hole
R4 kla k swallow, kla sand, c Chong, a j far
Syllables ending in -, - and -h are only attested with R1 et R3.
5. Reconstruction of initial plosives:
As in many languages of the Southeast Asia, Pearic dialects underwent a phonetic restructuring of
the initial plosives:
Low series (or series 2): devoicing of voiced initial plosives, generaly */b d g/> /p t c k/; in Chong
Klong Phlu /p t c k/. These shifts are associated with vowel raising and breathiness.
High series (or series 1): Comparison shows two types of treatment of voiceless initial plosives. In
the first type, the most simple initials remain unchanged, */p t c k/= /p t c k/ as in Khmer and in Thai. In the
second type, the most characterized, it is */p t c k/> /p t c k/. A vowel lowering can be associated with
this. An exception: Kompong Thom Pear, surrounded by Khmer speakers, only attests the simple shift.
According to the principle of the regularity of sound change, this two-ways treatment would be an
anomaly. Headley (1985) had adopted a provisional solution by reconstructing two series of plosives, */P T
C K/ for the simple type, and */p t c k/ for the characterized type. The state of knowledge at the time did not
permit a solution to the problem.
Proto Pearic Pearic dialects Kg Thom
Ferlus Headley 1985
p *p p p
*p
p *P p p
t *t t t
*t
t *T t t
c *c c c
*c
c *C c c
k *k k k
*k
k *K k k
Table 3: Proto-Pearic plosives according to Headley (1985)

In my notation, superscripts and are only used to differentiate the two types of correspondences.

Ferlus Headley examples
*p (p) *p poj [poj] after, pon [pon] four
*p (p) *P pa [pa] flower, kpat [kpat] cotton
*t (t) *t ta [ta] weave, tam [tam] crab
*t (t) *T tot [tot] head, ktak [ktak] tongue
*c (c) *c coj [coj] to plant, cin [cin] ripe
*c (c) *C cok [cok] run, cat [cat] knife
*k (k) *k ken [ken] son, kn [kn] mouse
*k (k) *K ka [ka] moon, kt [kt] low
The type */p t c k/> /p t c k/, named mutation germanique by Haudricourt (1965), is
sporadically attested in the Austroasiatic area. Apart from Pearic, it exists in Khasi and in Phay/Tin. Its
scarcity compared to the simple type could allow one to consider it the standard shift in Pearic. As for the
type */p t c k/= /p t c k/, very largely spread, it is attested in particular in Khmer and Thai, languages in
contact with Pearic. I think that this when this is the case in Pearic, it is due to the influence of Khmer. It is
known that in Khmer and Thai the phenomena of the restructuring of initials occurred in second half of the
seventeenth century. At this time the Pearic dialects would have formed a continuous territorial unit, except
for the isolated Kompong Thom Pear in which the marked process did not occur.
It is thus necessary to re-examine the historical phonetics of Pearic by taking account the influence
of the Khmer language.
6. Origin of final - in Pearic
Generally, final glottal stop - in Pearic does not correspond regularly to PMK - which is preserved
in Khmu, Waic, Mon (except *-i> today oa -ay) and partially in Vietic.
Chong register Khmu *Waic Mon
hand ti R1 ti te toa tay
louse ci R1 se si coa cay
pestle khi R1 cnre ri
fruit pli R1 ple pli
eight kti R1 snte
rattan se R1 r ri
rice field (waj3) s R1 re
grand father ta R1 ta ta
monkey wa R1 va hwa
eat ca R1 sa c ca
grass ktu R1 t
grandson co R1 s
dog c R1 s s
sesame ko R3 ra
leg plu R3 blu
forest, woods pri R3 bri bre
banyan tree (nem2) ci R3 ri sa jray
day ki R1 (<*R2) si se toa tay
ground, soil te R1 (<*R2) pte kt te ti
leaf la R1 (<*R2) la hla hla sla
stone, rock kmo R1 (<*R2) smo m tma
dream po R1 (<*R2) mpo rmo kp lpa
shell kl R1 (<*R2) kl nl knao knu
Michel Ferlus 43

<MM kinlo
paddy (haj1) k R1 (<*R2) rko rko
rain (n.) kma R3 (<*R4) kma m kma
thorn kla R3 (<*R4) crla hl
khala
Table 4: Distribution of final glottal stop
In the older transcriptions, the final - of Chong corresponds to a glottal constriction on the vowel
while giving an impression of dissyllabism. Let us compare some transcriptions of Suy Kompong Speu
(Baradat 1941) with those of Chong Klong Phlu (Siripen 2001).
Suy (Baradat) Chong (Siripen)
Ground, soil th- te
Leaf sla-a la
Stone, rock thma-a kmo
Acid, sour ch- co
Skin trl-a kalo

However, there are exceptions. It should not be forgotten that the former authors, in spite of their
merits, were not professional linguists. Historically, the final - in Pearic comes from a glottal constriction,
this is the reason why it does not correspond to PMK final glottal stop. Current - is only the result of a
recent secondary development. By writing the glottal constriction as --, one can propose the syllabic
evolution CVV>CV.
At this point, two important facts are highlighted: (i) Pearic - does not originate from PMK -, (ii)
Pearic - (CV) must be reinterpreted as a syllabic glottalization (creakiness) -- (CVV). As a consequence,
final - must be removed from Early Proto Pearic. The system of final plosives thus had only four units */p t
c k/.
In Chong (Siripen 2001), open syllables - and syllables in - (only short vowels) are attested with
only registers R1/R3. On the basis of preceding remarks, it is possible to reorganize the two sub systems
(syllables - and -) in a sole system while transferring syllabes ending in - from R1/R3 into R2/R4 on the
model of other final vowels (table 5). As a consequence, Pearic languages must be reconstructed whithout
final - as in Khmer (Ferlus 1992), in Katuic and in Bahnaric (Sidwell 1998, 2005).
R1 [CV] R1 [CV]
Syllables - and -
With two registers
R3 [CV ] R3 [CV ]

R1 [CV] R2 [CV]<[CVV]
Reorganization: syll. - with
four registers
R3 [CV ] R4 [CV ]<[CV V]
Table 5: Reorganization of Pearic registers by syllable types
Following this reorganization, the distribution of new open syllables in the four registers is of the
same type as that of the closed syllables (table 6).
R1 [CV] R2 [CVV]
Syllables -:
four registers
R3 [CV ] R4 [CV V]


R1 [CVC] R2 [CVC]
Closed syllables: four registres
R3 [CV C] R4 [CV C]
Table 6: Distribution of Pearic registers by syllable types
The most recent Proto Pearic (Late Proto Pearic), the stage preceding the devoicing of voiced
plosives initials, must be reconstructed with the contrast /creaky vs modal/. It will be explained later (see .8),
on the one hand, that this contrast does not come from the PMK, and on the other hand, that it can be
explained by an ancient syllabic contrast /tense vs lax/ (henceforth /T vs L/), possibly due to the influence of
Middle Chinese. At this stage, my hypothesis is based primarily on the geographical and temporal
coincidence that three Mon-Khmer groups lying along a known Chinese trade route developed similar /T vs
L/ contrasts.
7. Stages of the evolution from PMK to Pearic
It is now possible to present a chart showing the principal steps of the phonetic shifts of the Pearic
branch from the PMK stage up to modern times.
Stages
States
Shifts
(1) Proto Pearic Mon-Khmer Syllabes with final *-.
Loss of final *- > formation of syllables -
(2) Early Proto Pearic No syllable with final -.

Formation of /tense vs lax/ (influence of Middle Chinese) which
evolve to /creaky vs modal /. Note: Loss of final *- was also
propagated in Bahnaric, Katuic and Khmer. In Monic, it affected
only the rime *-i (oa -ay in modern Mon).
(3) Late Proto Pearic
No syllable with final -.
Marked register /creaky/.

Devoicing of plosives initials: */b d g/>/p t c k/ (in general).
Reinforcement of unvoiced plosives initials: */p t c k/>/p t c k /
(regular) or preservation as /p t c k/ (influence of Khmer).
Formation of register contrast /clear vs breathy/.
(4) Pearic at registral stage
No final -. Four registers system: /clear/,
/creaky/, /breathy/ and /breathy-creaky/.
Recreation of final - in tense syllables -.
(5) Current Pearic languages
New final -. Four registers system: /clear/,
/creaky/, /breathy/ and /breathy-creaky/.
Diversification by final consonants.
Table 7: from PMK to Pearic languages
8. Origin of creakiness
The creakiness is not very widespread in MK languages, it exists only in Vietic, in a part of Katuic
and in Pearic (Diffloth 1989). I already showed how a syllabic contrast /T vs L/ of Old Chinese had been
propagated in Vietic while modifying its phonation (Ferlus 2004). On this model, one can suppose that the
Michel Ferlus 45

contrast /creaky vs modal/ in Pearic can be also explained by this old syllabic contrast /T vs L/ of OC.
However, it is difficult to prove this categorically.
8.1 Formation of contrast /T vs L/ in Ancient Chinese.
The syllabic type in OC was (Cv)CV(C), a part of the vocabulary was made up of monosyllables
CV(C) the other part of sesqui-syllables CvCV(C). The coalescence of initials in sesqui-syllables developed a
tenseness /T/, while monosyllables became lax /L/. Thus syllabic contrast CvCV(C) vs CV(C) was coupled
with contrast /T vs L/. The evolution was continued by the monosyllabization and the formation of a syllabic
contrast CV(C)/T vs CV(C)/L in MC, associated with modifications of vocalic aperture, vowel lowering in
T-syllables and vowel raising in L-syllables (Ferlus 2009).
Old Chinese Middle Chinese
transfered to Vietic, Katuic and
Pearic
CvCV(C) (tenseness) CV(C)/T (v. lowering) T(ense)
CV(C) (laxness) CV(C)/L (v. raising) L(ax)
Table 8: Chinese Register Development
It will consider here only the contrast /T vs L/ which I will argue was propagated into Vietic, Katuic
and Pearic.
8.2 Transfer and evolution of contrast /T vs L/ in Vietic (Ferlus 2004)

Early Proto
Vietic syll.
Late Proto Vietic finals
*/p t c k/ */s h/ *//
*/m n
r l w j/
T CvCVC not affected not affected * > ()
glottalization
m n ()
L CVC not affected not affected unchanged () not affected ()
sc-nng hi-ng
ngang-huyn
sc-nng
sc-nng
ngang-huyn
Table 9: Vietic Register Development
The most outstanding fact of Vietic is the creation of open syllables in Late PVM. This fact created
conditions for the formation of the three fundamental tones represented by ngang-huyn, sc-nng and hi-
ng in Vietnamese. Of note: the feature /T/ is strong enough to cause the loss of final -, but not enough to
affect the voiceless final plosives. The voiced finals were glottalized and are represented by sc-nng tones
in Vietnamese.


8.3 Transfer and evolution of the contrast /T vs L/ in Katuic (Diffloth 1989)

Early Proto
Katuic syll.
Late Proto Katuic finals
*/p t c k/ */m n / */r l s h w j/ *
T CvCVC m n j m n r l s h w j v
L CVC p t c k m n r l s h w j
Table 10: Katuic Register Development
Contrast /T vs L/ has affected only some dialects (Katang, Talan, Yir/Ong) in the East of Katuic.
Other Katuic languages (Suoy, Kuy/Kuoy, S/Bru, ) were not affected. To simplify, I did not take account
of the vocalic length in the development of /T/. The effect of tenseness is more important in Katuic than in
Vietic.
8.4 Transfer and evolution of the contrast /T vs L/ in Pearic

Early Proto
Pearic syll.
Late Proto Pearic finals
*/p t c k/ */m n / */r l s w j/ */h/ *
T CvCVC p t c k m n r l s w j h v>v
L CVC p t c k m n r l s w j h
Table 11: Pearic Register Development
One can observe the re-creation of syllables -. Except for the syllables -h, all the others were
glottalized under the effect of the tenseness. The Pearic languages are those where the effects of tenseness are
generalized the most, but where the finals are the least corrupted. Exception, in Chung Yul, final plosives
were nasalized at creaky register: /p t c k/> /m n / while merging with nasals of the same register
(Isara 2009). This phonetic change is recent because it also affects the final /s>t/ (>/n/).
9. The Han trail and its Linguistic Implications
During the 3rd-8th centuries CE, Chinese texts reveal the existence of dependencies of the Chinese
Empire, located between the Middle Mekong and the north of present Cambodia.
The Records of the Three Kingdoms (sn gu zh ) record that to the 3rd century, a state
named Tang-ming (tng mng ), located north of present Cambodia, sent embassies to the emperor of
China (Pelliot 1903: 251). This practice indicates a nominal authority of China over this area.
In his great treatise of geography (sh do zh ), the author Kia Tan (Ja dn ), 8th
century, details the land route from the Chinese possession of Kiao-tche (jio zh ; Sino-Vietnamese:
Giao ch ), the present north of Vietnam, and leading to the dependency of Wen-tan (wn dn ) (Pelliot
1904: 210). It is thought that Wen-Tan was just one of the names of Tchen-la (zhn l ), in other words
Ancient Cambodia which extended farther north than present day Cambodia.
What would be the reason of the existence of these dependencies in an outlying region from China
and linked to Kiao-Tche by roads cut through geographical obstacles? It is clear that the roads described in
the texts were only those controlled by the Chinese, of the great transcontinental trade route connecting
southernmost China to the gulf of Thailand, and becoming a sea route toward India by a portage through the
Isthmus of Kra. This land route, a priori difficult, was essential to avoid the Cham who controlled the sea
route from China to India by the strait of Malacca. Tatsuo Hoshino (2002) remarkably studied the trans-
Michel Ferlus 47

Mekong route to the Wen-Tan, despite various difficulties of locating the places quoted by the Chinese
sources. We will call the part of the transcontinental trade route located between Kiao-Tche and the gulf of
Thailand the Han Trail.

Figure 1: A tentative map of the trans-peninsular trade route or Han trail,
leading from Kiao-Tche (ancient Vietnam) to the gulf of Thailand, and beyond to India.
What is the relation of the Pearic populations, now scattered in Cardamomes, with these trade route?
According to the ethnologist Marie Martin, the oral traditions of Samre mention a Chong kingdom before the
arrival of the Khmers. In addition, the Khmers of Chanthaburi had the memory of an old Chong capital
located on present Phnom Sebap (Martin 1997: 70). These places, located between the Great Lake and the
Gulf of Thailand, are the possible homeland of Pearic and a natural point of arrival of a trade route coming
from central Indochina.
I have argued above that the contrast /T vs L/ of Ancient Chinese (OC>MC) had been transferred
into Vietic, Katuic (partially) and Pearic. It can be objected that only a small part of Katuic attests this sound
change, but I think that it is due to the expansion of the West Katuic, which was not influenced by Chinese.
While at the same time the Tchen-la of Land
31
pushed back the East Katuic towards the margins (into the
hills east of the Mekong). Let us recall that the basic population of Land Tchen-la before the unification of
Cambodia was mainly the Bru (p lu ) ethnic group as documented by Chinese authors (Ferlus 2005).
However, these three linguistic groups are precisely located at the both ends and in the middle of this trade

31
The Chinese sources distinguished Tchen-la of Land and Tchen-la of Water, apparently refering to
Cambodia in-land from Cambodia around the Great Lake and lower Mekong.

route which during centuries was covered by Chinese travellers and traders. This coincidence between a
linguistic fact, formation of /T vs L/ in Vietic, Katuic and Pearic, and a trade route where the Chinese
carrying this contrast circulated, is sufficiently remarkable to deduce from it that this fact is not due randomly.
References
ASEMI: Asie du Sud-Est et Monde Insulindien
BEFEO: Bulletin de lcole Franaise dEtrme-Orient
BSEI: Bulletin de la Socit des Etudes Indochinoises
ILCRD: Institute of Languages and Culture for Rural Development
JSS: The Journal of the Siam Society
MKS: Mon-Khmer Studies

Antelme, Michel Rethy & Hlne Suppya Bru-Nut. 2001. Dictionnaire Franais-Khmer. Paris, L'Asiathque.
Baradat, R. 1941. Les dialectes des tribus smr. Manuscrit de lEcole Franaise dExtrme-Orient, Paris.
267p.
Baradat, R. 1941. Les s
mre ou Pe
r, population primitive de lOuest du Cambodge. BEFEO 41: 1-150.

Bastian, Adolf. 1868. Reise durch Kambodja nach Cochinchina, tome 3. Herman Gostenoble, Jena. [samreh
89-90; khamen boran 264-265; xong 295-298]
Blagden, Charles Otto. 1894. Early Indo-Chinese influence in the Malay Peinisula, as illustrated by some of
the dialects of the aboriginal tribes. Journal of the Straits Branch of the Royal Asiatic Society 27:
21-56.
Brengues, Docteur Jean. 1905. Notes sur les populations de la rgion des montagnes des Cardamomes. JSS
2(1): 19-47.
Cabaton, Antoine. 1905. Dix dialectes indochinois recueillis par prosper Odendhal. Etude linguistique par
Antoine Cabaton. Jounal Asiatique, 10
e
srie, 5: 265-344.
Crawfurd, John. 1828 [2000]. Journal of an Embassy from the Governor-General of India to the Courts of
Siam and Cochin China. London: Henri Colburn. [reprint 2000, New Delhi]
Diffloth, Grard. 1989. Proto-Austroasiatic Creaky Voice. MKS 15: 139-154.
Edmondson, Jerold A. 1996. Voice qualities and inverse filtering in Chong. MKS 26: 107-116.
Ellul, Jacques. 1968. Enregistrement magntique de vocabulaire saoch, rgion de Veal Renh, Cambodge.
Ferlus, Michel. 1981. A propos d'un allongement vocalique devant la finale r en khmer ancien. ASEMI 12(1-
2): 101-109.
Ferlus, Michel. 1992. Essai de phontique historique du khmer (Du milieu du premier millnaire de notre re
l'poque actuelle), MKS 21: 57-89.
Ferlus, Michel. 2004. The origin of tones in Viet-Muong. In Papers from the Eleventh Annual Meeting of the
Southeast Asian Linguistics Society 2001. Edited by Somsonge Burusphat. Arizona State University:
297-313.
Ferlus, Michel. 2005. L'intrt linguistique des transcriptions chinoises concernant le Cambodge ancien (Fou-
nan et Tchen-la). Dix-neuvimes Journes de Linguistique de lAsie Orientale, 30 juin - 1
er
juillet,
2005. Paris: CRLAO (EHESS-CNRS).
Ferlus, Michel. 2009. What were the four Divisions of Middle Chinese ? Diachronica 26(2): 184-213.
Filippi, Jean-Michel. 2000. Lexique et analyse phontique du Saoch. Manuscript.
Filippi, Jean-Michel. 2008. Recherches prliminaires sur les langues des minorits du Cambodge. Publi par
l'UNESCO et l'IDSSC, Phnom Penh, Editions Funan.
Garnier, Francis. 1873 [1976]. Voyage dexploration en Indo-Chine effectu pendant les annes 1866, 1867,
et 1868 Paris, Hachette. Vol. 2: 498-506.
Gregerson, Kenneth J. 1976. Tongue-root and register in Mon-Khmer. Austroasiatic Studies Part 1, Oceanic
Special Publication N 13. Edited by Philip N. Jenner, Laurence C. Thompson & Stanley Starosta.
The University Press of Hawaii: 323-369.
Gubler, J. s d. Liste de mots pears recueillis dans les Phnom Krevanh. Manuscript.
Guesdon, Pre Joseph. 1930. Dictionnaire cambodgien-franais, 2 vols. Paris: Plon.
Harmand, J. 1878-79. Notes de voyage en Indo-Chine: les Kouys Ponthey-Kakk. Annales dExtrme-
Orient 1: 322-339.
Michel Ferlus 49

Haudricourt, Andr-Georges. 1965. Les mutations consonantiques des occlusives initiales en Mon-Khmer.
Bulletin de la Socit de Linguistique de Paris 60(1): 160-172.
Headley, Robert K. 1977. A Pearic vocabulary. MKS 6: 69-149.
Headley, Robert K. 1978. An English-Pearic vocabulary. MKS 7: 61-94.
Headley, Robert K. 1985. Proto-Pearic and the classification of Pearic. In Southeast Asian Linguistic Studies
presented to Andr-G. Haudricourt. Edited by Surya Ratanakul, David Thomas, & Suwilai
Premsrirat. Bangkok, Mahidol University: 428-478.
Henderson, Eugnie J.A. 1952. The main features of Cambodian pronunciation. Bulletin of the School of
Oriental and African Studies 14(2): 149-174.
Hoshino, Tatsuo. 2002. Wen Dan and its Neighbours: The Central Mekong Valley in the Seventh and Eighth
Centuries. In Breaking new Grounds in Lao History, Essays on the Seventh to Twentieth Centuries.
Edited by Mayoury Ngaosrivathana and Kennon Breazeale. Chiang Mai, Silkworms Books.
Huffman, Franklin E. 1976. The register problems in fiftheen Mon-Khmer languages. Austroasiatic Studies
Part 1, Oceanic Special Publication N 13. Edited by Philip N. Jenner, Laurence C. Thompson &
Stanley Starosta. The University Press of Hawaii: 575-589.
Huffman, Franklin E. 1985. The Phonology of Chong, A Mon-Khmer Language of Thailand. In Southeast
Asian Linguistic Studies presented to Andr-G. Haudricourt. Edited by Surya Ratanakul, David
Thomas, & Suwilai Premsrirat. Bangkok, Mahidol University: 355-388.
Isara, Choosri. 2002. Dialects of Chong. MKS 32: 55-70.
Isara, Choosri. 2004. Chaak and other Chong place names: linguistics remains of a Mon-Khmer people in
Eastern Thailand. MKS 34: 67-77.
Isara, Choosri. 2007. Investigating Contact-induced Language Change: Cases of Chung (Saoch) in Thailand
and Cambodia. Thesis, Mahidol University.
Isara, Choosri. 2009. Chung (Saoch) of Thailand and Cambodia: phonological and lexical comparisions.
MKS 38: 69-85.
Isarangura, Nai Noe. 1935. Vocabulary of Chawng words collected in Krat Province. JSS 28: 173-186.
Jacob, Judith M. 1974. A Concise Cambodian-English Dictionary. London, Oxford University Press.
Kanika Kenikanon & al. 1979. Chong. Bangkok, Bannakij. [in Thai]
Kunwadee Phaetphithak. 1996. Phaasaa Chong Muu Baan Khlong Saeng, Tambon Dan Chumphon, Amphoe
Borai, Changwat Trat. MA Thesis. Silapakorn University. [A description of the Chong language in
Khlong Saeng village, Borrai district, Trat province]
[Leclre, Adhmar. ~1900]. Vocabulaire franais-sauch-cambodgien. Manuscrit de lEcole Franaise
dExtrme-Orient, Paris. 65p. [identified by the spelling sauch only used by Leclre, see BSEI 57,
1910]
Leclre, Adhmar. 1910. Les Sauch. BSEI 57: 91-114. La langue des Sauch, 112-114.
Malone, Dennis L. 2001. Language development in a minority language community: Report of the Chong
Writers Workshop. MKS 31: 141-152.
Martin, Marie A. 1974a. Remarques gnrales sur les dialectes pear. Asie du Sud-Est et Monde Insulindien
5(1): 25-37. Paris, Centre National de la Recherche Scientifique.
Martin, Marie A. 1974b. Esquisse phonologique du smree. ASEMI 5(1): 97-106. Paris, Centre National de la
Recherche Scientifique.
Martin, Marie A. 1975. Les dialectes Pear dans leurs rapports avec les langues nationales. JSS 63(2): 86-95.
Martin, Marie A. 1997. Les Khmers Daeum Khmers de l'origine. Socit montagnarde et exploitation de la
fort. De l'cologie l'histoire. Paris: Presses de l'EFEO.
Mntrier E. 1926. Monographie de la Circonscription rsidentielle de Kampot. Saigon: Editions dExtrme-
Orient. 89p. [saoch vocabulary p. 41-42]
Morizon, Ren. 1936. Essai sur le dialecte des populations Pears des Cardamomes. Paris, Editions Inter. 65p.
Moura, J. 1883. Le royaume du Cambodge. Paris, Ernest Leroux, 2 vols. [porr and samreh vocabulary, vol. 1:
440-447]
Noppawan Thongkham. 2003. The phonology of Kasong at Khlong Saeng Village, Danchumphon Sub-
District, Bo Rai District, Trat Province. MA Thesis, Mahidol University, Thailand.
Pannetier, Adrien Louis Marie. s. d. Notes sur les Pears, accompagn dun vocabulaire. Manuscript. Paris,
EFEO. [not easily usable]
Parkin, Robert. 1991. A Guide to Austroasiatic Speakers and Their Languages. Oceanic Linguistics Special
Publication N 23. Honolulu: University of Hawaii Press.
Pavie, Auguste. 1881-82. Excursion dans le Cambodge et le Royaume de Siam, III. Excursions et
Reconnaissances 4(10): 99-146. [voc. tchiong 114-116]

Pelliot, Paul. 1902. Mmoires sur les coutumes du Cambodge. BEFEO 2(2): 123-177. Traduction du texte de
Tcheou Ta-Kouan.
Pelliot, Paul. 1903. Le Fou-nan. BEFEO 3(2): 248-303.
Pelliot, Paul. 1904. Deux itinraires de la Chine en Inde la fin du VIIIe sicle. BEFEO 4: 131-413.
Pelliot, Paul. 1951. Mmoires sur les coutumes du Cambodge. uvres posthumes de P. Pelliot. Paris: Adrien
Maisonneuve. Traduction revue et augmente de 1902.
Pinnow, Heinz-Jrgen. 1959. Versuch einer historischen lautlehre der Kharia-Sprache. Wiesbaden, Otto
Harrassowitz.
Pornsawan Ploykaew. 2001. Samre grammar. Thesis. ILCRD, Mahidol University.
Pornsawan Ploykaew. 2001. The phonology of Samre. MKS 31: 15-27.
Pornsawan Ploykaew. 2005. The study of Samres world view: the belief in sorcerer spirits. MKS 35:
171-182.
Purtle, Dale. 1970. Manuscript vocabulary of Chong collected at Veal Renh, Kampot Province.
Saifon Luamkhan. 1991. Phaasaa Chong Muu Baan Nam Khun 1, Khlong Phlu, Amphur Makham, Changwat
Chanthaburi. MA thesis. Silapakorn University.
Schmidt, Pater Wilhelm. 1907, 1908. Les peuples mon-khmer, trait dunion entre les peuples de lAsie
centrale et de lAustronsie. BEFEO 7 (1907): 213-263 & BEFEO 8 (1908): 1-35.
Schmitt, J. 1904. Les chongs. Revue Indochinoise II(3): 161-162.
Seidenfaden, E. 1943. Review of R. Baradat, Les s
mre ou Pe
r, population primitive de louest du

Cambodge, BEFEO 41: 1-150, 1941. JSS 34(1): 73-80.
Sidwell, Paul. 1998. A Reconstruction of Proto-Bahnaric. Thesis, University of Melbourne.
Sidwell, Paul. 2005. The Katuic Languages, Classification, Reconstruction and Comparative Lexicon.
Lincom Europa.
Sirikarn Charoentham. 1987. Phaasaa Chong Muu Baan Thung Ta-In, Changwat Chanthaburi [Chong
Language, Thung Ta-In village, Chanthaburi province]. Thesis. Silapakorn University.
Siripen Ungsitipoonporn. 2001. A phonological comparision between Khlong Phlu Chong and
Wangkraphrae Chong. MA thesis. ILCRD, Mahidol University.
Skeat, Walter William & Charles Otto Blagden. 1906. Pagan Races of the Malay Peninsula, two vols.
London: Franck Cass & Co. Reprint 1966, New York: Barnes & Noble. [vocabularies from Doudart
de Lagre & Francis Garnier 1873: 498-517, and Moura 1883: 440-447]
Sunee Kamnuansin. 2002. The syntax of an endangered language: A case study of Kasong. Paper presented
to The 8th International Conference on Thai Studies (January 9-12, 2002). Nakhorn Pathom,
Thailand.
Sunee Kamnuansin. 2002. Kasong syntax. MA thesis. ILCRD, Mahidol University.
Sunee Kamnuansin. 2003. Syntactic characteristics of Kasong: An endangered language of Thailand. MKS 33:
167-182.
Surekha Suphanphaiboon. 1982. The phonological system of Chong language, Muban Takhianthong,
Tambon Takhianthong, Amphoe Makham, Changwat Chanthaburi. M.A. Thesis, Srinakharinwirot
Prasarnmit University, Bangkok. [in Thai]
Surekha Suphanphaiboon. 1987. Kan Suksa Klum Chatiphan Chao Chong. Research paper presented to The
Social Science Association of Thailand.
Suwaphat Chanmekha. 2003. The Hidden Language: a case Study of Kasong Language Attitudes, Uses and
Habilities in three Villages in Trat Province, Thailand. Thesis, Mahidol University. [in Thai]
Suwilai Premsrirat. 2001. Austroasiatic languages and peoples. In Annotated Bibliography for Ethnic
Minorities in Thailand N2. Nakhon Pathom, ILCRD, Mahidol University. [unpublished ms in Thai]
Suwilai Premsrirat. 2004. Ethnolinguistic Maps of Thailand. Ministry of Culture and Mahidol University.
Suwilai Premsrirat. 2008. Chong Dictionary. Center for Revitalization of Endangered languages, Institute of
Language and Culture for Rural Development.
Tcheou Ta-Kuan. 1295/1296. See Paul Pelliot 1902.
Theraphan L. Thongkum. 1984. Samre language. Journal of Thammasat University 15(1): 16-128. [in Thai]
Theraphan L. Thongkum. 1988. Phonation types in Mon-Khmer languages. In Vocal Physiology: Voice
Production, Mechanisms and Function. Edited by Osama Fujimura: 319-333.
Theraphan L. Thongkum. 1991. An Instrumental Study of Chong Register. In Austroasiatic Languages.
Essays in honour of H.L. Shorto. Edited by J.H.C.S. Davidson. School of Oriental and African
Studies, University of London: 141-160.
Thomas, David D. & Robert K. Headley. 1970. More on Mon-Khmer subgroupings. Lingua 25: 398-418.
Michel Ferlus 51

Ueda, H. & Okada, T. 2003. Saoch vocabulary. Reports on minority languages in mainland Southeast Asia.
Webber, Karl E. 1976. Ethnographic notes of the Chong population in Chanthaburi province southeast
Thailand. Thailand Research Project Diffusion of Innovation Coast Hinterland - Continuum
Preliminary. Office of Senior Research Fellow of Southeast Asia Institute of Heidelberg University
of German Culture.

Ghosh, Arun. 2011. Goal-marking in Munda with special reference to the Lower Munda languages. In Sophana
Srichampa and Paul Sidwell (eds.) Austroasiatic Studies: papers from ICAAL4. Mon-Khmer Studies Journal Special
Issue No. 2. Dallas, SIL International; Salaya, Mahidol University; Canberra, Pacific Linguistics. pp.52-68.
52
Goal-marking in Munda with special reference
to the Lower Munda languages
Arun Ghosh
University of Burdwan
Introduction
Goal is used here to subsume different semantic roles, viz. recipient (R), direction (D), location (L),
beneficiary (B), causee (C) and experiencer (E) to test how these are marked and whether these are marked
alike in the Munda languages. It has been argued in typological literature that the goal-marking morphemes
are extremely elastic in nature. These morphemes are mostly polysemous in character as they encompass
different semantic roles.(Rice and Kabata 2007:451) This is true synchronically as well as diachronically.
Here in identifying different goal-marking morphemes and their extension pathways emphasis is given on the
three south Munda languages, Remo, Gutob and Gta which are classified as Lower Munda by S.
Bhattacharya (1975) on some morphosyntactic criteria like a. lack of object and dative control in the verb; b.
lack of marker for dual number; and c. use of genitive marker different from Upper Munda
(Bhattacharya:1975), represented by Santali, Mundari, Ho, Korku, Kharia, Juang, Sora, Pareng and a host of
minor Munda languages like Turi, Asuri, Birhor etc. There are two ways of encoding the patient (P) in Lower
Munda: prefixation and suffixation depending on selection between the participant noun and pronoun.
Marking also depends on animacy and selection between human and non-human participants.
In Remo prefixation is employed regardless of the participants being noun or pronoun. In Gutob and
Gta selection of the process depends on the selection between participant noun or pronoun-- prefixation is
used when the participant P is pronoun and suffixation is used when the participant P is noun. In case of
marking of the theme (T) the inanimate T is zero marked in all the three languages and if animate it receives
the same marking as P. In Remo the recipient (R) receives the same marking regardless of its being noun or
pronoun. In Gutob and Gta it receives the same marking as P. The prominal R in Gta can also get double
marking, that is, prefix and suffix both. Remo marks the direction (D), location (L), experiencer (E), causee
(C) and beneficiary (B) with the same prefix assigned for P and R. In Gutob pronominal D, L, E, C and B are
marked by the same prefix assigned for pronominal P and R and the nominal D, L, E, C and B are marked by
the same suffix assigned for nominal P and R. Gta marks E and C with the same prefix or suffix assigned for
P and R but uses different markers for D, L, and B. As L, D, C, E and B in Remo and Gutob, and E and C in
Gta are assigned the same marker as R all are treated here as a single argument, viz. Goal. The dative-
accusative marker performing the role of P and R is treated here as the basic signal and there from extension
pathays are tried to be located. After discussing the goal-marking in Lower Munda languages the Upper
Munda languages are taken up for consideration to see whether the same extension pathways hold in the
whole Munda group.
Theoretical Background
The idea of locating the extension pathways of the semantic roles stems from different readings like
Blansitt (1988), Heine (1990), Rice and Kabata (2007), Malchukov, Haspelmath and Comrie (2007) and
Kittil (2008). To substantiate his Function Contiguity Hypothesis Blansitt located in the available data of
Arun Ghosh 53

Munduruku a common function marker for dative, allative and locative.(Blansitt 1988:180) His Function
Conguity Hypohesis was schematically represented as
Object- Dative- Allative- Locative (Ibid. 177)
or
Object = Dative = Allative = Locative (Rice & Kabata 2007: 463)
He generalized among others if an adposition occurs as both object marker and allative marker, it
also occurs as dative marker. If an adposition occurs as both dative marker and locative marker it also occurs
as allative marker. (Blansitt 1988:186)
Heine (1990) in his model of Dative extension observed that the non-cognate suffixes -ke and -ro of
the two Nilo-Saharan languages, Ik and Kanuri respectively, shared many of the same functions, including
the marking of indirect objects, directional locatives, goals, benefactives, purposes, reasons, manner and time
complements, as well as marking subordinate clauses and serving as a derivational suffix to mark adverbs
(Heine 1990: 129).
32
Rice & Kabata argued that direction markers tend to extend to recipient marking
diachronically.
33
In their endeavour the main purpose was to investigate the concomitant semantic roles and
functions that the principal goal-marking morpheme in a language also marks, such as location, recipient,
possessor, experiencer, purpose, etc., as well as more traditionally conceived morphological cases, such as
dative, genitive, etc. (Ibid.: 452) Contesting Blansitt's Function Contiguity Hypothesis (1988) and Heine's
Dative Extension model they proposed a third model of the grammaticalization pattern of the Allative
(2007:494) in which they proposed four principal semantic domains, SOCIAL, SPATIO-TEMPORAL,
LOGICAL-CONTEXTUAL and MENTAL with Social leading to Recipient extending to Addressee and
Benefactive, Spatio-temporal to Locative extending to Time and Ablative, Logical-contextual to Purposive
extending to Reason and Infinitival and with Mental to Conceptual extending to Perceiver and Experiencer.
34

Malchukov, Haspelmath and Comrie while looking at the ditransitive constructions across languages found
that verbs of physical transfer like give, sell, lend etc. describe a scene in which an agent participant
causes an object to pass in the possession of animate receiver (= recipient). They also found that verbs
denoting mental transfer like show or tell behave alike. That is, verbs having recipient and verbs denoting
direction behave alike. In their view the animate argument of show' and tell' is not a recipient in the
narrow sense, but can be regarded as an R-argument. Moreover they also found that benefactive
constructions having prospective recipient are expressed like the ditransitive constructions. In their findings
causee in the causative constructions is found to be functioning like the R, which led them to argue that this
is of course not an accident, because the meanings of transfer verbs contain a cause element. (Malchukov,
Haspelmath and Comrie 2007: 2-3) Kittil is in favour of neutralizing the differences between the semantic
roles of recipient and goal in his discussion on differential goal marking (DRM). Observing that both
recipient and goal can be seen as end points of transfer (Kittil 2008: 248) he is in favour of lumping these
roles together. He further argues that there are languages which accord R a uniform marking regardless of
animacy, which also justifies seeing these as different manifestations of a single semantic role. (Kittil 2008:
248)
Taking a cue from previous discussions and having a close look at the Munda data, especially those
of the Lower Munda
35
languages, it has been proposed that in Munda all the three arguments, viz. recipient
(R), direction (D) and beneficiary (B) can be considered as manifestations of a single semantic role, that is,
goal, symbolized as R by Kittil. Aside these three, causee, permisee, person permitted for receiving

32
Sally Rice and Kaori Kabata, Cross-linguistic Grammaticalization Patterns of the ALLATIVE. Linguistic
Typology, 11, 2007, p.464.
33
See also Seppo Kittila & Silvia Luraghis Questionnaire for their on-going project Differential marking of
spatial relations: the case of direction with human landmarks. (E-mail circulation).
34
For a schematic presentation of Rice and Kabata's model see Sally and Kabata, 2007: 494, Figure 21.
35
The name Lower Munda was given to three extreme south Munda languages, Remo, Gutob and Gta by S.
Bhattacharya in his article titled Munda studies: A new classification of Munda (Indo-Iranian Journal, v.xvii: 1, pp.97-
101) where he isolated these languages from other Munda languages on some morphosyntactic criteria.
54 Goal Marking in Munda

something or undertaking some action and the Experiencer
36
are also encoded in the same way as the R in
Munda. Other functions are also taken into account as causee involves some kind of transfer of action from
one person to the other; person permitted is the receiver of the permission and in the psychomatic verbs
whether sensory, emotional or physical like feel good, be angry, be hungry the logical subject of the
sentence receives the R-marking. Moreover, as the roles discussed so far involve some kind of transfer of
action and/or event we are also in favour of considering these roles together into a single role, designated
hereafter as goal (R). In the following discussion we will primarily concentrate on the encoding of goal (R) in
the Lower Munda langages, viz. Remo, Gutob and Gta and compare it with other Munda languages,
especially Santali, Mundari, Korku, Kharia, Juang , Sora and Birhor.
Goal-marking in Lower Munda
The lower Munda languages, Remo, Gutob and Gta although taken together, do not encode goal
in a uniform manner. The extension of semantic role of R varies to some extent. The following examples will
illustrate how the R is encoded in the three Lower Munda languages, Remo, Gutob and Gta. Let us take up
Remo
37
data first:
Remo:
1. gitin -ni muj gim b-
3SG ACC/DAT-1SG one goat give-PST
'He gave me a goat.

2. -n ni sur b-ti-
ACC/DAT-2SG 1SG medicine give-NPST-1SG
'I shall give you medicine.

3. rmo -sonj muj gim b-
Ramu ACC/DAT-Sonya one goat give-PST
'Ramu gave Sonya a goat.

4. i -meri muj upor b-
Hadi ACC/DAT-Mary one gift give-PST
'Hadi gave a gift to Mary.

In the examples the R whether nominal or pronominal is uniformly marked with the ACC/DAT
prefix -. All the sentences are in ditransitive frame with a human R and non-human T. The non-human T is
unmarked. Examples (3) and (4) have alternative forms in which the Rs Mary and Sonya can receive
double marking, that is, accusative-dative prefix - and address clitic
38
-l, (which can also be dispensed with)
as in the examples (5) and (6):
5. i -meri-l sur b-
Hadi ACC/DAT-Mary-CLIT medicine give-PST
Hadi gave medicine to Mary.

6. rmo -meri-l muj upor b-
Ramu ACC/DAT-Mary-CLIT one gift give-PST
Ramu gave a gift to Mary.

36
Subject receiving R-marking has been designated as Experiencer by Gregory D.S. Anderson (2008) and
Leukas Neukom (2000) and as Dative subject by Manideepa Pattanayak (2008).
37
Remo is the language of the Bondas distributed in and around the Bonda Hills in the Khairput
block of the Malkangiri district of Orissa. There are at least two known dialects of the languagePlains
Remo and Hill Remo. The people call their language remsam. Data presented here represent the Hill dialect.
38
The clitic is basically used as an endearing element with the person addressed or referred. It is
used with the subject noun (+human) as in sobit-l mj-n klm sun b- Sabita sold her pen. selnl
ni-pulj ru- The girl brought water for me. It can also be dispensed with without disturbing the
meaning or reference, as in sobit-la mj-n klm sun b- and seln ni-pulj ru-.
Arun Ghosh 55

The goals of the verbs of mental transfer like show, tell, bring and motion verbs like throw
receive the same encoding as the R. Examples (7)-(11) illustrate the argument:
7. gitin -ni usrm u- bib
3SG ACC/DAT-1SG story tell-PST RED-give
s-gi-t
come-PRF-NPST
He has come to tell me a story.

8. gitin -ni n-
ru- b-
3SG ACC/DAT-1SG 1-child bring-PST give-PST
He brought and gave me a child.

9. -j-j sun-
ACC/DAT-mother-PB.KIN tell-PST
He told his mother'.

10. gitin -ni muj bire tul-
3SG ACC/DAT-1SG One stone throw-PST
He threw me a stone.

11. gitin -ni muj bl knel-
3SG ACC/DAT-1SG One ball throw at-PST
He threw a ball at me.

In the examples from (7)-(11) the direction is encoded through the same marking as the R with the
human goal. The goal of the motion verbs like come, enter, climb, go is encoded in the same way,
that is with the accusative-dative prefix -, as in the examples (12-15):
12. -io wiy-
ACC/DAT-house go-IMP
Go home. (Frank Fernandez 1967: 112)

13. ni ()- io-bo
39
gy-ti-
1SG ACC/DAT-house enter-NPST-1SG
I enter/ shall enter the house. (Frank Fernandez 1967: 68)

14. my sk-set -semu yk-t
3SG come-PART ACC/DAT-tree climb-NPST
Having come he will climb the tree. (Sahu, Samantaray, Patel 1993:30)

15. -ni l
ACC/DAT-1SG come
Come to me.

In contrast to Gutob and Gta (see examples 50, 63-64) Remo marks the place-names and locations
with the same accusative-dative marker, as in (16)-(17):
16. ni -mundlip uj-ti-
1SG ACC/DAT-Mudulipada go-NPST-1SG
I shall go to Mudulipada.

39
Here R-marking morpheme - is optional. Similarly -b can also be dispensed with in favour of
-.

17. n - uj-tu-n
2SG ACC/DAT-weekly market go-NPST-2SG
You will go to the market.

There is another way of encoding the D/L (direction/location) in Remo, that is, through the D/L-
marking suffix -b and that applies to both human and non-human goals. Consider examples (18)-(20):
18. br-b uj- mri gim-se ru
market-D/L go-IMP and goat-meat bring
Go to the market and bring mutton.

19. ktr-b uj- -p sur b-t
doctor-D/L go-IMP who ACC/DAT-2PL medicine give-NPST
Go to the doctor who will give you medicine.

20. gitin l-b uj-g
3SG field-D/L go-PRF
He went to the field.

21. bdl i-b uj-s
-seg mmr-seg sun-

Badal Hadi-D/L go-CLIT-PART something-PART tell-PST
Going to Hadi Badal told something'.

In these examples the direction-location (D/L) marker -b encodes both human and non-human
goals. In (18) and (20) -b encodes the non-human goal and in (19) and (21) the human goal. Regular
alternation of this suffix with the accusative-dative prefix as found marking the R clearly shows that R and
D/L in Remo are treated alike. Examples (22)-(25) justify our argument:
22. ni io-bo gy-t-i
1SG house-D/L enter-NPST-1SG
VS
ni -io gy-t-i
1SG ACC/DAT-house enter-NPST-1SG
I enter the house. (Frank Fernandez 1967: 68)

23. oom guso-bitre i-t
arrow dog-in COP-NPST
VS
oom -guso i-t
arrow ACC/DAT-dog COP-NPST
The arrow is in the dog. (Frank Fernandez 1967: 68)

24. ni kori-bgbo lyk-t-i
1SG chair-D/L sit-NPST-1SG
VS
ni -kori ljk-t-i
1SG ACC/DAT-chair sit-NPST-1SG
I sit on the chair. (Frank Fernandez 1967: 68)

25. gulj-ne kon-b onj-g
boys-PL hill-D/L go-PRF
VS
gulj-ne -kon onj-g
boys-PL ACC/DAT-hill go-PRF
The boys went to the hill.

Arun Ghosh 57

The psychosomatic verbs
40
encode the subject as experiencer of transfer point of bodily or mental
sensations with the same R-marker, that is -, as in the examples (26)-28). It is the experiencer subject which
acts as the transfer point of physical or psychological experience and not the experincer object, quite in line
with the argument put forward by Croft that the experiencer-subject verbs are purely stative and that the
experiencer is characterized as simply being in a mental state regarding the stimulus. (Palmer 1994:27;
Croft 1991:214-15). The experiencer subject here, however, does not control agreement in the verb.
26. -ni kuru-sg ni ki sum-o-ni
ACC/DAT-1SG hungry-PART 1SG rice eat-PST-1SG
As I was hungry I ate rice.

27. -ni lu-lor-usu-gu-t
ACC/DAT-1SG RED-vomit-DESID-PRF-NPST
I wish to vomit.

28. -ni si susp-en-t
ACC/DAT-1SG fever RED-get-PROG-NPST
I am getting fever.

Beneficiary which prototypically refers to entities, usually animates that are indirectly affected by
the action of the verb' (Palmer 1994:31) can optionally be encoded with the R marker, that is, the beneficiary
in the benefactive constructions may be marked with the accusative-dative - or may be marked with -
along with the clitic pulj for', as in (29)- (31):

29. sobit ni-pulj muj klm sb-
Sabita 1SG-CLIT one pen buy-PST
Sabita bought a pen for me.

29a. sobit -ni muj klm sb-
Sabita ACC/DAT-1SG one pen buy-PST

29b. sobit -ni-pulj muj klm sb-
Sabita ACC/DAT-CLIT one pen buy-PST

30. seln ni-pulj ru-
3SG 1SG-CLIT water bring-PST
She brought water for me.

30a. seln -ni ru-
3SG ACC/DAT-1SG water bring-PST
She brought water for me.

31. gitin ni-pulj semu sisep'-en-t
3SG 1SG-CLIT tree RED-cut-PROG-NPST
He is cutting the tree for me.

31a. gitin -ni semu sisep'-en-t
3SG ACC/DAT-1SG tree RED-cut-PROG-NPST

40
For a detailed discussion of the psychosomatic verbs in Mundari see Osada (1992: 104-109).

31b. gitin -ni-pulj semu sisep'-en-t
3SG ACC/DAT-CLIT tree RED-cut-PROG-NPST

The causee in the causative constructions receives the same R-marking as the cause element also
involves transfer of action from one person to the other. In Remo any or all of the verbs can be transformed
into a causative regardless of transitivity with the simple meaning of causing someone to perform the
relevant action'. (Palmer 1994: 215). Here also C receives R-marking. Consider (32) (33):
32. rmo -sonj-l ki -sum-
Ramu ACC/DAT-Sonya-CLIT rice CAUS-eat-PST
Ramu caused Sonya to eat rice.

33. - -gu -lem
ACC/DAT-child-PB.OFF CAUS-sleep
Make the child sleep.

The extension pathways of R in Remo may be shown schematically as follows:

ACC-DAT--->RECIPIENT=DIRECTION=LOCATION=EXPERIENCE =BENEFICIARY = CAUSEE

A closely related language and a close neighbour of Remo Gutob
41
shows more or less the same
typological pattern as found in Remo. Here too the R in the ditransitive frame, direction-location of the
motion verb as also the Experiencer of the psychomatic verbs are encoded in the same way. While the
pronominal goal is marked by the accusative-dative prefix o- the nominal goal is marked by the clitic
lj/pulj. The D/L-marker in respect of both human and non-human goals is bo which optionally alternates
with the R-marker lj/pulj, meaning for, that is, the nominal R of the ditransitive frame optionally acts as
goal. The beneficiary, the prospective or projected recipient is also found to be receiving the accusative-
dative prefix o- along with the clitic lj/pulj in case of pronominal goal and the clitic lj in case of
nominal goals . Consider (34) (38) where the pronominal goal (both recipient and goal) in the ditransitive
frame receives the accusative-dative prefix o-:
34. djkori o-ni be-tu
kindly ACC/DAT-1SG give-NPST
Kindly give me.

35. pen o-ni kodli be-tu-pen-ki
2PL ACC/DAT-1SG banana give-NPST-2PL-Q
Will you give me banana?'

36. mj o-ni sun-
3SG ACC/DAT-1SG tell-PST
He told me.

37. mj o-ni muj-ro bl ti-
3SG ACC/DAT-1SG one-CLAS ball throw-PST
He threw me a ball.

38. sobit o-ni muj-ro boj ri-
Sabita ACC/ DAT-1SG one-CLAS book bring-PST
Sabita brought me a book.

41
Gutob is the language of the Munda speaking Gadabas distributed mainly in the Lamtaput block of the
Koraput district of Orissa. Data analyzed here were collected in the Lamtaput block.
Arun Ghosh 59

Here the pronominal goals (R) of the verbs of physical and mental transfer are marked by the
accusative-dative prefix o- and the T is marked zero. Examples of the nominal goals are the following:
39. i meri-lj muj-ro upor b-
Hadi Mary-CLIT one-CLAS gift give-PST
Hadi gave a gift to Mary.

40. bdl sobit-nu-laj muj-ro boj ri-
Badal Sabita-GEN-CLIT one-CLAS book bring-PST
Badal brought a book to Sabita.

lj-pulj is sometimes found to be alternating with -bo as in the alternative sentence in (41a).
Compare (41) and (41a):
41. sobit sonj-nu-pulj muj-ro boj ri-
Sabita Sonya-GEN-CLIT one-CLAS book bring-PST
Sabita brought a book to Sonya.

41a. sobit sonj-nu-bo muj-ro boj ri-
Sabita Sonya-GEN-D/L one-CLAS book bring-PST
Sabita brought a book to Sonya.

The experiencer (E) is encoded with the accusative-dative marker o- in case of pronominal
experiencer and -lj in case of nominal experiencer. As in Remo the experiencer does not control agreement
in the verb. Consider (42) and (43):
42. o-ni sos lgj-guni
ACC/DAT-1SG thirst be affected-COP
I am thirsty.

43. o-ni besi duk-e-guni
ACC/DAT-1SG very be sad-PROG-COP
I am very sad.

Beneficiary is also marked by the accusative-dative marker o- and the clitic
42
lj. Consider (44) and
(45):
44. o-ni-laj ri-tu-nom-ki
ACC/DAT-1SG-CLIT water bring-NPST-2SG-Q
Will you bring water for me. (Asha Kiran Society 2002:14)

45. nisni mpru-pulj gisi ri-nen
Nisani god-CLIT fowl bring-3PL
Let them bring fowl for the Nisani God.

In the ditransitive frame with two argumentsR and T and both are human R is marked by -bo and
T is marked by -laj/pulaj. In similar constructions with R and nonhuman T the R is encoded with lj and the
T is left unmarked. Consider (46) and (47):
46. ni oon-lj jo-ej-nu-bo be-o-ni
1SG child-ACC/DAT mother-P.B-KIN-GEN-LOC give-PST-1SG
I gave the child to the mother.

42
laj is assigned the accusative status as it is basically used for P-marking.

47. rmu sonj-lj muj-ro gime be-o
Ramu Sonya-CLIT one-CLAS goat give-PST
Ramu gave Sonya a goat. (Rajan & Rajan 2001: 32)

The causee is encoded with the same accusative-dative marker as in (48) and (49):
48. o-nom ni ni-nu pu-ly b-soi-tu-ni
ACC/DAT-2SG 1SG 1SG-GEN father-CLIT CAUS-see-NPST-1SG
I will show my father to you. (Bhattacharya 1975: 164 )

49. nej gutob-log shebo-pulj samo b-gir-nej
2PL Gadaba-people British-CLIT language CAUS-learn-2PL
u-tu
AUX-NPST
We the Gadabas are teaching the Saheb (our) language.

Unlike Remo Gutob does not encode place-names with the R-marker. Consider (50):
50. ni bier mudlipa i-lom-ni
1SG tomorrow Mudulipada go-FUT-1SG
I shall go to Mudulipada tomorrow.

The extension pathways of R in Gutob as found in the examples (34)-(50) can therefore be
schematized as follows:
ACC-DAT -->RECIPIENT=DIRECTION=LOCATION=EXPERIENCER BENEFICIARY = CAUSEE

The other member of the group Gta
43
follows more or less the same pattern. Like Remo and Gutob
there are two markers for encoding R, one for pronoun and the other for noun. While the nominal R is
marked by -ke
44
the pronominal R is marked by the accusative-dative prefix -. Sometimes the pronominal
goal receives double marking, that is, pronominal R is marked by the accusative-dative prefix - and the
suffix -ke. The causee and the experiencer receive the same encoding, - in case of pronouns and -ke in case
of nouns. Pronominal goal with the verbs of physical transfer, mental transfer as also with motion verbs is
marked with the accusative-dative prefix -. Consider (51) (54):
51. -ni muj bi-l
ACC/DAT-1SG one rupee give-IMP
Give me one rupee.

52. m -ni husr ni-te
3SG ACC/DAT-1SG story tell-PST
He told me a story.

Nominal goal with the same verbs is marked with -ke:

43
Gta is the language of the Didayis spoken in the Kudumulguma block of the Malkangiri district of Orissa.
They themselves call their language Gtasa. There are three known dialects of the languagePlains Gta, Hill Gta and
Cut off area Gta. My data are drawn from the cut-off area. S. Bhattacharya calls the language by their ethno-name
Didayi.
44
ke may be a loan from Indo-Aryan as most of the eastern Indo-Aryan languages have this suffix used in
dative-accusative. But if it is a loan from Indo-Aryan then through which language does it come? The most plausible
answer would be Oriya, as it is the next neighbour to Oriya. But interestingly Oriya possesses the suffix in the form of
ku, not ke which is found in Bangla. And Bangla is in the neighbourhood of Gt in the southern part of Malkangiri
district where more than 600,000 Bengalees are staying in the Dandakaranya project area
Arun Ghosh 61

53. rmu sonj-ke muj gimi bi-te
Ramu Sonya-ACC/DAT one goat give-PST
Ramu gave a goat to Sonya.

54. n omru-ke muj cii bi-n-ge
1SG Domru-ACC/DAT one letter send-1SG-PST
I sent a letter to Domru.

Like Remo and Gutob causee also receives the accusative-dative encoding. Consider (55)-(56):

55. rmu jo-ke -o-te
Ramu baby-ACC/DAT CAUS-eat-PST
Ramu fed the baby.

56. m -n mi -o-te
3SG ACC/DAT-2SG sweets CAUS-eat-PST
He caused you to eat sweets.

Experiencer as perceiver of physical and mental sensation is encoded with the accusative-dative
marker as in (57)-(58):

57. -ni besikee sob-i-te
ACC/DAT-1SG very behungry-PROG-NPST
I am very hungry.

58. tn kito remw-ke bwe-tur-l remw-ke g-ke
that god man-ACC/DAT spit-DS man-ACC/DAT itch-PST
The god spit on the man and the man started to feel itchy.
(Anderson 2008:697)

Example (58) is typical in the sense that along with the experiencer subject (the second italicized
field in the sentence) the experiencer object (P) is marked with -ke (the first italicized field in the sentence).
Examples (59) and (60) show double marking of R:

59. -me-ke
45
ter gisi bi
ACC/DAT-3SG-ACC/DAT this fowl give
Give him this fowl. (Bhattacharya 1975b:166)

60. -m-ke bi-l
ACC/DAT-3SG-DAT give-IMP
Give him. (Panda: 1989:13)

In a ditransitive frame with R and T and both human, R is marked with ke and the T is left
unmarked as in (61) and in the similar frame with human R and non-human T the R is marked with ke and
the T is left unmarked as in (62):

61. n j-ke muj muom bi-te
1SG mother-DAT one child give-PST
I gave the child to the mother.

62. n j-ke mui gisi bi-te
1SG mother-DAT one fowl give-PST
I gave mother a fowl.

45
As examples (59) and (60) are taken from Bhattacharya (1975) and Pandas (1994) data no pattern can be
found regarding the distribution of -, -ke and -ke.

Unlike Remo and like Gutob Gta does not overtly mark the place-names. Compare (63) and (64)
with (65):

Gta:

63. ni njige mudulip wj-e
1SG tomorrow Mudulipada go-FUT

Gutob:

64. ni bijer mudlip i-lom-ni
1SG tomorrow Mudulipada go-FUT-1SG

Remo:

65. ni -munlip uj-ti-
1SG ACC/DAT-Mudulipada go-NPST-1SG
I shall go to Mudulipada.

In Gta there is one locative-directional marker -rini which is not found to mark the R although
there is enough evidence of R marking the direction.

The extension pathways of R in Gta may be schematised as follows:

ACC-DAT----> RECIPIENT = DIRECTION = CAUSEE = EXPERIENCER

The pattern of extension of the semantic role of R as discussed so far is not the same in all the three
Lower Munda languages. While it encompasses the roles of R, D, L, E, B, C in Remo and Gutob, it covers
only R, D, C and E in Gta, Beneficiary and Location falling outside the purview of R. Now let us consider
the grammaticalization pattern of the R in other Munda languages outside Lower Munda.
Goal-marking in Upper Munda:
The grammaticalization pattern of R as attested in other Munda languages, can be shown in (66)
(86):

Santali:
The recipient (66), direction (67-69) and causee (70) are marked by the applicative and pronominal
incorporation in the verb in Santali:

66. dk m--d-i--e
rice give-A-TM-1SG-FIN-3SG
He gave me rice.

67. hc--d-i--e
come-A-TM-1SG-FIN-3SG
He came to me. (SK cf. LN. 49)

68. uni li--ko--e
3SG tell-A-3PL-FIN-3SG
He will tell them.

69. dre- bgt--k-kn-th
kn-
tree-1SG look at-A-INAN-COP-COP:PST-FIN
I was looking at the tree. (Bo. Cf. LN.53)
Arun Ghosh 63

70. l-oco--d-e--e
see-CAUS-A-TM-3SG-FM-3SG
He showed it to him.

Mundari:
The same process holds in Mundari as in (71-73):

71. mi set-ko- om--d-ko-
food dog-PL-1Sg give-A-TM-3PL-FIN
I gave food to the dogs.

72. dsi-ko-e kji--t-ko-
servant-PL-3SG tell-A-TM-3Pl-FIN
He told his servants. (Bhattacharya 1975: 150)

73. diku- itu--d-ko-
Hindu-1SG teach-A-TM-3PL-FIN
I have taught Hindi to them. (Osada 1992: 95)

Korku:
In Korku the recipient as in (74), direction as in (75) and causee as in (76) are marked the same
suffix ke:

74. am-ke ij mya-kama:y ghaliba
you-to(IO) I one-work-obj give up
I will give you a work. (Nagaraja1999: 46)

75. ij Dic-ke ambesasa kule-c-lakken
I he-obj. mango-bring send-per-cont
I am sending him to bring mango. (Nagaraja1999:46-47)

76. dij' koro-ke i-en ghleij
that man-ACC/DAT 1SG-D/L show
Show me that man. (Bhattacharya 1975:157)

Birh:
In Birhor the recipient receives the ke marking along with pronominal incorporation in the verb, as
in (77):

77. oni sim i-ke ago-i-mi
that fowl 1SG-ACC/DAT bring-1SG-IMP
Bring that child to me. (Bhattacharya 1975: 154)

Kharia:
In Kharia the recipient as in (78), causee as in (79) and direction as ((80) are marked by the suffix
te:

78. m m- be-nom-te i-te tere
2SG 2SG-GEN son-2SG-ACC/DAT 1SG-ACC/DAT give
You give me your son. (Bhattacharya 1975:158)

79. i i- p-i-te m-te ob-iyo-ij'
1SG 1SG-GEN father-1SG-ACC 2SG-ACC CAUS-see-1SG
I will show my father to you. (Bhattacharya 1975: 158)


80. i ulg-ho dru-te ep-ni-g
1SG leaf-for tree-ACC/DAT climb-1SG-FUT
I will climb the tree for leaves. (Bhattacharya 1975: 159)

Juang:
In Juang te marks the recipient (81), causee (82)-(83) direction (84):

81. rm m-te r i bbnd kete
Ram Shyam-ACC/DAT Self House burn about
me-gt-yr gm-
NEG-say-INF say-PST
Ram told Shyam not to tell anybody about his burning house.
(Pattanayak 2008:212)

82. i-te ju-k gt-ro-ki b-so-i
1SG-ACC/DAT Juang-GEN language-DEF-PL CAUS-teach-1SG
Teach me the language of Juang. (Bhattacharya1975:160)

83. m i-te goble-kon-om-te m-am-o-i
2SG 1SG-ACC/DAT nephew-2SG-ACC/DAT CAUS-see-1SG
You will show your nephew to me. (Bhattacharya1975: 160)

84. rm- [i- inj-te ngr-te]
Ram-GEN [1SG-GEN house-ACC/DAT go-INF ]
bel -si-n
time NEG-be-PST
Ram does not have time to go to my house. (Pattanayak 2008: 70)

Sora:
Like the Lower Munda languages Sora also markes the recipient with the prefix -, as in (85)-(86):

85. kuni -trb-ji -mr tiy
that ACC/DAT-flower-PL ACC/DAT-man give
Give those flowers to the man. (Bhattacharya 1975:162)

86. kuni -trba-ji -ni-ji tiy
that ACC/DAT-flower-PL ACC/DAT-3PL give
Give those flowers to them. (Bhattacharya 1975: 162)

Examples cited can be sub-grouped into two: examples (66) to (73) and (77), (82)-(83) on the one
hand and (74) to (76), (78)-(79) and (85)-(86) on the other. They differ in respect of R-marking in the verb.
In Santali (66-70) Mundari (71-73), Birh (77) and Juang (82-83) the R is cross-referenced on the verb. But
while Santali and Mundari take the applicative - tagged on to the pronominal referent or to the TAM their
close kin Birh does not take any applicative marker, simple pronominal forms are added to the verbal
predicate in the form of infix. In Korku (74)(76), Kharia (78-79), Sora (85-86) and in Juang (81) there is no
verbal cross-referencing. While in Kharia (78-80) and Juang (81 and 84) the R is marked by te in the
sentence Sora (85-86) marks R by the accusative-dative prefix -. One interesting point about the second
group (Kharia and Sora) is that the suffix te and prefix - employed for marking the R is also used for
marking the T which may lead to ambiguity. To disambiguate the phenomenon the ordering of T and R may
be taken into account. Kharia and Sora normally follow the S T R V pattern, though S R T V pattern for
Kharia can not be ruled out.
46
Although supporting evidences are not available for all languages examples
like (67), (70), (72), (77), (74), and (80) clearly show that recipient and direction/location converge and
support our hypothesis that movement or transfer in respect of object or location can be subsumed under R

46
This is confirmed by John Peterson in an informal talk.
Arun Ghosh 65

(goal) in Munda. In some languages outside Lower Munda Beneficiary (B) is also marked in the same way as
R. Consider (87)(90):
Santali:

87. prkm bel--ko-m pe-ko hc-en-
Bedstead place-A-3PL-2SG visitor-PL come-PST:INTR- FIN
Place a bedstead for them to sit on, visitors have come. (Bo.cf. LN: 48)

Mundari:
88. dru-m mg---t-n-
tree-2SG cut-A-1SG-AM-ITM-FIN
You are cutting the tree for me. (Osada 1992: 93)

Korku:
89. Do Di-ku-ke co:ja-ma lija sege-ba De-Ten
and dem-pl-IO why-re cloth bring-np dem-abl
i-ke-ka mya soy-lubu sa:li
I-IO-rest one good-cloth bring-imp
And why do you bring new clothes for them rather bring one new cloth for me.
(Nagaraja 1999:155)

Juang:
90. r judi gl-b n-de i-te
3SG If Keonjhar-DIR go-CON 1SG-ACC/DAT
pcol m-R-r-ene
shawl FUT-bring-FUT
If he goes to Keonjhar he will bring a shawl for me. (Pattanayak2008: 36)

At least in these three languages outside Lower Munda Beneficiary receives the R- marking.

The Experiencer also receives the R-marking outside Lower Munda. Consider (91)-(92) and (95):

Santali:

91. hol-e dk--t-le-
yesterday-3SG rain-A-TM-1PL-FIN
Yesterday we had rain. [lit. rain affected us] (Bo. Cf. LN: 50)

Mundari:

92. sown-j---
smell-AM-TM-1SG-FIN
I have sensed a smell. (Osada 1992: 106)

Korku:

93. i-en kapara kasu-lakken
I-dlc head ache-cont
I am having headache. (Nagaraja 1999: 100)

94. i-en da dadam-en
I-dlc water thirsty-pt
I am thirsty. (Nagaraja1999:100)


Juang:

95. i-te eml i-sere
1SG-ACC/DAT cold be-PRF
I have got cold. (Pattanayak 2008: 193)

Here in these examples as the syntactic subject is affected by either the event (rain as in Santali) or
sensation (smell as in Mundari) it receives the R-marking. In Korku, however, the experiencer (E) receives
a separate D/L marker which is not used for marking the recipient, causee or beneficiary. Along with
direction, beneficiary, causee and experiencer applicative also encodes the person permitted or receiving
permission in permissive construction in Santali as in (96):
Santali:

96. kiri-oco--d-i--e
buy-C/P-A-TM-1SG-FIN-3SG
He permitted me to buy.

In the permissive construction (96) the argument receiving permission is also encoded with the
applicative - as the permissee is also an indirect recipient.
The grammaticalization pattern of R in the Upper Munda languages can be schematically
represented in two blocs---- one represented by Santali and Mundari where R is marked by the applicative
and the other represented by Korku, Birhor, Kharia, Juang and Sora where R is marked by the accusative-
dative morpheme.
1. APPLICATIVE[Santali and Mundari]--->
RECIPIENT = DIRECTION = CAUSEE = BENEFICIARY =EXPERIENCER = (PERMISSEE)
47

2. ACC/DAT [Korku, Birhor, Kharia, Juang and Sora]---> RECIPIENT=DIRECTION= CAUSEE=
BENEFICIARY=(EXPERIENCER)
48

On the basis of the discussion made separately of Lower and Upper Munda so far the over all
semantic-functional extension of R in Munda may be delineated in a linear model as follows:
R RECIPIENT=DIRECTION=LOCATION=CAUSEE=(BENEFICIARY)=EXPERIENCER
=(PERMISSEE)

The model explicates that all the roles are not shared by all the Munda languages. The common
extension is Recipient, Direction, Location, Cause and Experience. But while Gutob, Remo, Santali, Mundari,
Korku, Juang have Benefactive extension other languages do not share this feature (as attested in the data).
The Permissive construction is found only in Santali. Of the Lower Munda languages Remo and Gutob have
elaborate extension process covering six rolesR, D, L, E, B, C uniformly. Gta shows only four-way
extension covering R, D, C, E. Outside Lower Munda Santali has elaborate pattern covering six rolesR, D,
C, B, E, P. For paucity of data nothing conclusive can be said about Birhor and Sora, though the general
tendency leads us to the assumption that the six-way extension path of R might have been the general
tendency of proto-Munda.
Conclusion
The paper examines encoding of goal (R) from a Munda perspective. The phenomenon comprises
different types: one, whether the recipient of a ditransitive construction receives the same encoding as the
goals of direction-location. It has been found from the available data that recipient and directional-locational

47
The permissive construction is available in Santali, hence PERMISSEE is kept in paranthesis.
48
Experiencer is found to be receiving R-marking in Kharia and Juang. Korku marks E with a direction-
location marker which is not used to mark the recipient.
Arun Ghosh 67

goals receive the same encoding, not only in Lower Munda but in other Munda languages as well. Two,
whether beneficiary (B), causee (C) and permisee (P) receive the same marking as R. It has been found that
all these roles receive the same marking as R. Three, whether coding of the syntactic subject coincides with R.
It is also found to be partially true. The present study (to the best of my knowledge) is the first cross-Munda
analysis of R (goal), raising many issues that deserve to be investigated in more detail. It is true that it may
not be so easy to make generalization about R-marking in Munda in the light of the data presented. Some
more data need to be incorporated for that. Still it is to be noted that the phenomena discussed give some idea
about the general tendency of R-marking in the Munda languages.
Abbreviations
A applicative INAN inanimate
ABL ablative INF infinitive
ACC accusative IO indirect object
AKS Asha Kiran Society ITM intransitive marker
AM aspect marker LN Leukas Neukom
AUX auxiliary NEG negative
B beneficiary/benefactive NPST non-past
Bo. P.O.Bodding OBJ. object
CAUS causative P patient
CLAS classifier PART Participle
CLIT clitic PB.KIN post base kinship marker
C/P causative/ permissive PB.OFF post base marking offspring
CON conditional PL plural
cont continuous PROG progressive
COP copula PRF perfect
D/L direction/location PST past
DAT dative Q question marker
DEF Definitive R recipient (goal)
DESI desiderative RED reduplicated
DIR directional marker R&R Rajan and Rajan
DL dual number SG singular
dlc direction-location SK L.O. Skrefsrud
FIN finite marker T theme
FUT future TAM tense aspect marker
GEN genitive TM tense marker
IMP imperative VC verb clitic.

Acknowledgement
Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany has provided financial
assistance for undertaking field survey in the southern part of Orissa for collecting data in Remo, Gutob and
Gta. I would like to thank Professor Bernard Comrie, Director, Department of Linguistics, Max Planck
Institute for Evolutionary Anthropology, Leipzig, Germany for arranging the financial assistance and for
constant support in undertaking linguistic study of the dying Munda languages. I also owe much to him for
going through the first two drafts of this paper. I owe much to Dr Seppo Kittil of the University of Helsinki
for his constant encouragement in building the theoretical framework of this paper and for going through the
first draft of the paper. I thank Dr. Nicholas Evans of Australian National University and Dr. John Peterson
of the University of Leipzig for their comments on earlier drafts of the paper. I also remember with gratitude
the help I received from my informants, especially Lachimi Challan, Soma Kirsani, Badal Dhangramajhi,
Jaya Bisoyi, Subhas Muduli, Jogu Bodnayek, Guru Kirsani and Niranjan Sisa during my field trip to Koraput
and Malkangiri in July-August 2009, February-March and September-October 2010. Lacunae if there is any
is of course mine.
References
Anderson, Gregory D S and K. David Harrison, 2008. Remo. In: Self (ed.) The Munda Languages. London:
Routledge, pp. 557-632.
Asha Kiran Society, 2002. Gutob Language Learners Guide. Lamtaput, Koraput.
Bhattacharya, Sudhibhushan, 1975. Studies in Comparative Munda Linguistics. Simla: Indian Institute of
Advanced Studies.
_____. 1975. Munda studies: A new classification of Munda. Indo-Iranian Journal, v. XVII: 1, pp.97-101.
Blansitt, Edward L. Jr., 1988. Datives and Allatives. In: Michael Hammond, Judith Moravcsik & Jesica
Wirth (eds.), Studies in Syntactic Typology. pp.173-191. Cambridge: CUP.
Bodding, P.O. 1929 Materials for a Santali Grammar, Part II. Dumka: The Santal Mission of the Northern
Churches.
_____. 1925-1929 Santal Folktales. Oslo: H. Aschehoug & Co.
_____. 1929-1936 A Santal Dictionary. Oslo: A. W. Broggers Boktrykkeri.
Croft, W. 1991. Syntactic Categories and grammatical relations. Cambridge: Cambridge University Press.
Fernandez, Frank, 1967. A Grammatical Sketch of Remo: A Munda Language. PhD thesis, University of
North Carolina at Chapel Hill.
Kittil, Seppo, 2008. Animacy effects on differential Goal marking. Linguistic Typology, 12 pp.245-268.
Lukas Neukom, 2000. Argument marking in Santali. Mon-Khmer Studies, 30, pp.95-113.
Malchukov, Andrej, Martin Haspelmath, Bernard comrie, 2007. Ditransitive constructions: A typological
overview. First draft.
Matson, D. M., 1964. A Grammatical Sketch of Juang. PhD dissertation, Madison, University of Wisconsin.
Nagaraja, K. S. Korku Language Grammar, Text, And Vocabulary. Tokyo: Institute for the Study of
Languages and Cultures of Asia and Africa, Tokyo University of Foreign Studies.
Osada, Toshiki, 1992. A Reference Grammar of Mundari.Tokyo: Tokyo University of Foreign Studies.
Panda, Pramodakumari, 1989. Didayi. Bhubaneswar: Academy of Tribal Dialects and Culture.
Pattanayak, M., 2008. Juang. In: Gregory D S Anderson (ed.), The Munda Languages. London: Routledge,
pp. 508-556.
Rajan, H. & Jamuna Rajan, 2001. Grammar write up of Gutob-Gadaba. Lamtaput: Asha Kiran Society.
Rice, S. & K. Kabata, 2007. Crosslinguistic grammaticalization patterns of the ALLATIVE. Linguistic
Typology 11, pp.451-514.
Sahu, Samantaray and Patel, 1993. Bonda. Bhubaneswar: Academy of Tribal Dialects and culture.
Skrefsrud, L.O. 1873. A Grammar of the Santhal Language. Benares: Medical Hall Press.

Theraphan L-Thongkum, Chommanad Intajamornrak, Kanitha Putthasatien, Yupaporn Huadsiri, and Supaporn Phalipat .
2011. A Synopsis of Mal Phonetics. In Sophana Srichampa and Paul Sidwell (eds.) Austroasiatic Studies: papers from
ICAAL4. Mon-Khmer Studies Journal Special Issue No. 2. Dallas, SIL International; Salaya, Mahidol University;
Canberra, Pacific Linguistics. pp.69-80.
Copyright vested in the authors.
69
A Synopsis of Mal Phonetics
Theraphan L-Thongkum, Chommanad Intajamornrak,
Kanitha Putthasatien, Yupaporn Huadsiri, and Supaporn Phalipat
Department of Linguistics, Chulalongkorn University
Abstract
The consonants, vowels and pitches or tones of Mal (Tin), a Khmuic language of the
Mon-Khmer language family, were acoustically studied using computer programmes and
were statistically tested. The three places of articulation of initial stops in Mal can be
identified using the Locus Equation (LE) method. Even though the distinct slopes and y-
intercepts of the linear equation can characterize the place of articulation, it is more
effective to use the slope values. The ratio of duration for long-to-short vowels is 2:1. The
overall vowel space of short vowels is smaller than that of their long counterparts but the
variation within the space of each short vowel is more dispersed. The phonation type of the
initial sonorant has an influence on the F0 value of the following vowel. Some Mal varieties
have become two-tone languages, /high/ and /low/. The five stages of tonal evolution
induced by language contact with Khammueang are postulated.
1. Introduction
The Lua or Tin are a Mon-Khmer speaking ethnic group living in Nan province, northern Thailand
and the adjacent Sayaburi province of Lao PDR (see map in the appendix). In Nan province, the 34,600 Lua
speakers are scattered throughout ten districts: Bo Kluea, Pua, Chaloem Prakhiat, Chiang Klang, Thung
Chang, Santisuk, Mueang Nan, Mae Charim, Wiang Sa and Song Khwae (more detailed information can be
found in L-Thongkum and Intajamornrak, 2009). According to Filbeck (1972), the Lua language of Nan
province consists of two major dialects, i.e. Mal and Pray, and each of these two dialects has a few sub-
dialects or varieties. This classification has been confirmed by Singnoi (1988) and Jirananthanaporn (1993).
However, the Lua have never identified themselves as Mal or Pray. Perhaps, the ethnonyms and language
names Mal and Pray are only known to anthropologists and linguists, not among the Lua or the local
people of Nan province. They always say we are Lua people and we prefer to be called Luaby outsiders.
They do not want to be KhonTin (meaning natives or local inhabitants) because it has derogatory
connotations, i.e. stupid, ignorant and uncivilized people.
In Nan, besides the Mal-Pray language, the other Mon-Khmer languages are also spoken, i.e. Khmu
and Mlabri. A number of cognate words can be found in Lua (Mal and Pray), Khmu and Mlabri, such as to
snap (the fingers): *phlas (Proto-Tin), phlayh (Mal), phat (Pray), pl
jh (Khmu Rawk) and pllh (Mlabri);

to wake (someone up): *phrh (Proto-Tin), phyh (Mal), phrh (Pray), phr
h (Khmu Rawk) and paph

(Mlabri).
49
These Mon-Khmer languages belong to the Khmuic branch of the Mon-Khmer language family
(Filbeck, 1972 and 1978).

49
These two examples are from L-Thongkum (2007). More information on the Mal, Pray, Khmu and Mlabri
lexicon, about 2,400 entries, is provided in L-Thongkum et al., for further comparative studies.
70 A Synopsis of Mal Phonetics

Mal has two major varieties, tonal and non-tonal. The tonal variety, spoken in Ta Luang and
Yotdoiwatthana villages in the districts of Pua and Bo Kluea respectively, was selected for our acoustic study.
The result of our literature survey shows that an acoustic study of Mal does not exist, even though Mal
phonology has been studied synchronically and diachronically by Filbeck (1972, 1976, 1978, 1990); Singnoi
(1988); L-Thongkum (2007) and L-Thongkum et al. (2007). Our acoustic studies of Mal relating to
consonants, vowels and tones were made under the research project Linguistic Diversity in Nan Province: A
Foundation for Tourism Development funded by the Thailand Research Fund (TRF) for three years (2004-
2007).
This paper is a synthesis of the research findings presented in three M.A. theses (Huadsiri, 2007;
Phalipat, 2007 and Putthasatien, 2007) and a Ph.D. dissertation (Intajamornrak, 2009c). The objectives of our
research on Mal phonetics are: to test whether the Locus Equation (LE) method can satisfactorily identify the
places of articulation of initial stops; to analyse and compare the duration, formant frequency and vowel
space of short and long vowels; and to investigate the acoustic characteristics of the two tones in order to
postulate the stages of tonal evolution from the past to the future.
2. Phonological sketch of Yotdoiwatthana Mal
Mal has many initial consonant clusters due to the reduction of prefixes, infixes and sesqui-syllables
(minor or unstressed syllables before major or stressed syllables), otherwise it can be said that the Mal
phonological system is rather simple in comparison with those of its sister languages. The consonant, vowel
and tone systems of Yotdoiwatthana Mal are shown below.
Consonants
Labial Alveolar Palatal Velar Glottal
Plosive Vl. unaspirated
Vl. Aspirated
Vd.
p
ph
b
t
th
d
c

k
kh

Nasal Vl.
Vd.
hm
m
hn
n
h

h

Fricative Vl. s h
Aproximant Vl.
Vd.
Pregottalised
hw
w
w
hl
l

j

Consonant clusters
Cw phw thw kw khw lw
Cl pl phl bl kl khl
Cj pj phj bj kj j mj sj
NC (l/j) mph mpl mphl nth ns k kh kl kj

Vowels
Front Central Back
Short Long Short Long Short Long
High i i u u
Mid e e o o
Monophthongs
Low a a

i () u Diphthongs
i e a a o i u
50

Tones (high) (low)

50
Proto-Tin (Proto-Lua) *-r has become *- in Yotdoiwatthana Mal, for example, *phar > ph to fly; *thar >
th rope, cord, string; *khe:r > kh: carambola and so forth.
Theraphan L-Thongkum et al. 71

3. Methodology
The data for the acoustic study of consonants, vowels and pitches or tones was collected at Ban Ta
Luang in Amphoe Pua which is a district of the province of Nan, northern Thailand. All of the words used in
the devised word lists were pronounced by three female speakers (30-50 years old) and directly recorded with
a high-quality microphone onto a computer notebook using Adobe Audition. With regard to the study of
pitch patterns which can reflex tonal evolution in Mal, three more Mal varieties spoken in Ban Kwet, Ban
Phu Kok, Ban Yotdoiwatthana and a conservative variety of non-tonal Pray spoken in Ban Huai Lom were
acoustically analysed and compared. An acoustic study of tones in the Pua variety of Khammueang
pronounced by native speakers and bilingual speakers in Mal and Khammueang was done to help confirm
our claims on the birth of tone in Mal.
In order to test whether the Locus Equation (LE) method can satisfactorily identify the three places
of articulation of initial stops in Mal, i.e. bilabial, alveolar and velar, suitable word lists were devised. The 81
words used in this study had CVV(C) syllable structure, initial stops with the three places of articulation and
three manners of articulation: p ph b t th d k kh and nine vowels: i e a u o . The pronunciation
of each speaker was recorded three times. The 729 test tokens (81 words x 3 times x 3 speakers) were
acoustically analysed with Praat version 4.4.13. The second formant frequency (F2) at the burst of the
obstructive airstream (F2 at burst) and the steady state of F2 at the 50% of vowel duration (F2 vowel) were
measured in hertz. The locus equation was calculated from a linear equation and then plotted on a linear
regression graph. The F2 values at the burst of the nine vowels were plotted along the y axis, while the F2
values in the steady state of the nine vowels were plotted along the x axis, in order to obtain the distinct
slopes and y-intercepts of the linear equation which can characterise the three places of articulation. See the
results in 4.1.
With regard to the acoustic analysis of vowels, nine short and long counterparts (18 vowels): i-i e-
e - - - a-a u-u o-o and -, were used. The duration and formant frequencies (F1, F2) of the
nine pairs of vowels were analysed with Praat version 4.5.06 and statistically tested with a t-test (<0.05). The
devised word list consists of 90 words, five words for each vowel. Three recordings were made for each of
the three speakers. The 810 test tokens (90 words x 3 times x 3 speakers) were acoustically measured. The
average values of F1 and F2, S.D., the vowel spaces and square units of the two sets of vowels (short vs. long)
and the vowel space of each vowel showing its variation were calculated and then plotted using the three
programmes: extractFeatures, Vowel plot and Polygon, written for us by Patthawi Chanwaiwit. See the
results in 4.2.
To test the hypotheses that the fundamental frequencies (F0) of vowels following voiceless initial
sonorants (hm hl) are higher than those following voiced initial sonorants (m l) and that the F0 difference is
statistically significant, a suitable word list for the tests was devised. The data from each of the three female
speakers was recorded three times. The 720 test tokens (80 words x 3 times x 3 speakers) were acoustically
measured with Praat version 4.4.04 and the significance of F0 difference was tested with a t-test (<0.05). The
investigation of F0 behaviour was carried out in two steps. In the first step, the acoustic characteristics of the
two tones (high vs. low) were analysed in order to obtain an overview of the tone shapes, then, the analysis of
F0 at every 25% (0%-100%) of the normalised duration was done. In the second step, which was the main
objective of our tone investigation, the same 720 test tokens mentioned previously were used again for
investigating the fundamental frequencies of vowels following voiceless and voiced initial sonorants. The
results of the measurement in real time at every 25 millisecond (msec.) of vowel duration were plotted with
Microsoft Excel. See the results in 4.3.
To see clearly how the Yotdoiwatthana variety of Mal has become tonal, the pitch or F0 patterns
occurring in native words having different types of initial consonant, final consonant and vowel height in the
other three Mal varieties and a Pray variety were also acoustically studied. The pitch patterns of Tai
(Khammueang) loanwords and those of native words (minimal pairs) were also analysed in order to test the
hypothesis that tonal evolution in Mal was induced by an external factor, i.e. language contact with
Khammueang spoken by the majority of people living in Nan province. Three word lists consisting of 265,
262 and 140 words were devised to suit the multipurposes of our acoustic investigation. For each purpose,
three female speakers were used and three recordings for each speaker were made using Adobe Audition

version 2. All together, the F0 values of 6,003 test tokens were acoustically analysed with Praat version
4.5.24 and the significance of the findings was tested with t-test (<0.05). See the results in 4.4.
4. Results
With regard to the acoustic characteristics of Mal consonants, vowels and tones, the following are
our findings.
4.1 Consonant
The Locus Equation (LE), a method for identifying the places of articulation of initial stops, used by
Sussman and Shore (1996) and Modarresi et al., (2005) was applied. The co-articulation of nine initial stops,
i.e. the bilabials /p ph b/, alveolars /t th d/ andvelars /k kh / and nine long vowels, i.e. /i e a u o /
in Mal was investigated by means of the LE technique.
The second formant frequency at the burst of vowels (F2 at burst) and in the steady state of the
vowels (F2 vowel) was measured and analysed.
The LE values were calculated from a linear equation plotted on a linear regression graph. The
values of F2 at burst are shown in the y axis and those of the F2 vowel are shown in the x axis, see examples
in Figure 1.


Figure 1 The locus equations of /b d /

The distinct slopes and y-intercepts of the linear equation can characterise the places of articulation
of the three sets of initial stops as shown in Figure 2. However, using the slope value seems to be more
effective in identifying the place of articulation than the y-intercept value. The results confirm the hypothesis
that the velar stop has the highest slope, while the lowest slope occurs in the alveolar stop. It is also affirmed
by the F-test that the difference in the slope values is statistically significant. More details can be found in
Huadsiri (2007, 2009).
0.34
0.13
0.97
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Bilabial Alveolar Velar
Place of articulation
S
l
o
p
e

v
a
l
u
e
s

Figure 2 The mean slope values of initial stops
4.2 Vowel
Mal has nine pairs of short and long vowels: /i-i:/, /e-e:/, /-:/, /-:/, /-:/, /a-a:/, /u-u:/, /o-o:/ and
/-:/. (Details can be found in the phonological sketch section.) The duration and formant frequencies (F1
and F2) of these eighteen vowels were acoustically analysed. The average durations of short and long vowels
are 212.05 msec. and 482.15 msec. respectively. Broadly speaking, the ratio of duration of long-to-short
vowels is about 2:1. The difference between the average duration of short and long vowels is statistically
significant. See Figure 3.

0
100
200
300
400
500
600
short long
D
u
r
a
t
i
o
n

(
m
s
e
c
.
)

Figure 3 The average durations of the short and long vowels

Generally, the short front vowels /i e / have a higher F1 but a lower F2 than those of their long
counterparts /i e /. This means that the short front vowels are lower and more centralised. For central
vowels, // is lower than // since // has a higher F1, whereas // and /a/ are higher and more centralised than
// and /a/ due to the fact that // and /a/ have a lower F1 but a higher F2. With regard to back vowels, the
short vowels /u o / have a higher F1 than their long counterparts /u o /. This indicates that the short
vowels are lower than the long vowels and that // is more peripheral than //, see Figure 4. The difference
between the F1 and F2 of the short vowels and their long counterparts is statistically significant, except for
the F1 of the pairs /-/ and /a-a/, and the F2 of the pairs /-/, /-/ and /a-a/.
i
e
a
u
o
i:
e:
:
:
:
a:
u:
o:
:
400
600
800
1000
1200
500 1000 1500 2000 2500 3000
F2 (Hertz)
F
1

(
H
e
r
t
z
)
Short
Long

Figure 4 The overall vowel spaces of the short and long vowels

The overall space of the short vowel is smaller (394, 914.92 square units) than that of the long vowel
(438, 329.98 square units) as can be seen in Figure 4. As is evident in Figures 5 and 6, each of the short
vowels in Mal has a greater variation within its space than that of the long vowels. More details can be found
in Phalipat (2007, 2009).

Figure 5 The vowel spaces of the short vowels

Figure 6 The vowel spaces of the long vowels

4.3 Tone
The acoustic characteristics of the two tones, i.e. /High/ and /Low/, were analysed. The fundamental
frequency at every 25% (0%-100%) of normalised times was measured. The result of the F0 measurements
indicates that the high tone has two phonetic shapes [high-falling] in the non-checked syllable and [high-level]
in the checked one, while the low tone has only one characteristic, i.e. [low-rising] in both types of syllable
as is shown in Figure 7.

Figure 7 The phonetic characteristics of the two tones

Filbeck (1978) classifies the Lua (Tin) language into three groups, i.e. A, B and C, using the
criteria of lexical and phonological development. He says that Mal B has two tones, the rising tone and the
non-rising tone. Some minimal pairs can be found in Filbeck (1972), for example, /kan/ work (loanword) -
/kaan/ defeated (loanword); /ca/ be able (loanword) - /caa/ to hire (loanword) and so forth. With
regard to pitch difference, there are three levels of pitch height: high, mid and low. These three pitch levels
correlate with the three degrees of stress: primary stress with high pitch, secondary stress with mid pitch and
unstressed with low pitch. There are three kinds of pitch contours: level, falling and rising. The high pitch
with falling contour (high-falling) and the mid pitch with falling contour (mid-falling) occur in smooth or
non-checked syllables having primary stress and secondary stress, respectively. The high-level pitch occurs
in long checked syllables with primary stress, while the mid-level pitch occurs in those with secondary stress.
Short checked syllables and syllables having a short vowel with a final /-h/ can have one of the two level
pitches, high-level or low-level.
Singnoi (1988) and Jirananthanaporn (1993) agree with Filbeck (1972 and 1978) and confirm that
Mal has two phonemic pitches or two tones, the rising tone and the non-rising tone. Singnoi (1988)
summarises the shapes of the two tones as is shown in Table 1.

Syllable structure Phonetic characteristics
Open and closed smooth syllable High-falling, Low rising
Checked syllable
Short
Long

Level
Level, Low-rising
Syllable with short vowel and /-h/ Level
Table 1 The phonetic characteristics of the Mal tones
(Adapted from Singnoi, 1988)


Our acoustic findings, to some extent, agree with Singnois description of the phonetic
characteristics of the two phonemic pitches of the Mal variety spoken in Ban Wangsao in Amphoe Chiang
Klang which is her research site. It is possible that in the Mal variety of Ban Yotdoiwatthana studied by our
research team and the one described by Singnoi (1988), even though both of them have two tones, the tone
shapes or the phonetic realisation of the two tones may be different. Moreover, since the two studies were
done about twenty years apart, phonological variation and change would not be abnormal.
4.4 Tonogenesis
Proto-Mal, reconstructed by Filbeck (1978), is non-tonal and most of the modern varieties of Mal
are still non-tonal. What is the cause of tone birth in Mal?
The Lua (Tin) or Mal-Pray were forced to move from the mountains and to settle in Pa Klang
Refugee Camp located in the lowland area of Amphoe Pua during the communist infiltration about forty
years ago. Later, due to a positive situation in the remote areas of Nan, they returned to the mountains. They
became bilingual in Mal and Khammueang during their stay in the lowlands.
Filbeck (1972) states that the two processes of tonal development, i.e. by means of independent
innovation and by means of contact with a tonal language (Thai) occurred in Mal (p.111). A number of
minimal and near minimal pairs of Thai loanwords found in Mal suggests that the Mal tonal system (rising vs.
non-rising) emerged through contact with Thai (p.115). However, the emergence of the rising tone seems to
have been the result of an independent innovation (p.116).
In our opinion, the two tones in Mal should be analysed as high vs. low, not falling vs. rising or non-
rising vs. rising. Khammueang has six tones: low rising (A1-2), mid-rising (A3-4), mid-level (B1-3), mid-
falling (B4), high-level (C1-3) and high-falling (C4). The Mal two tones (high vs. low) are assigned to
Khammueang loanwords, no matter what tones they are in the donor language. Khammueang words with
mid-rising, high-level and high-falling tones, when borrowed into Mal, tend to have the Mal high tone and
vice versa, the ones with low-rising, mid-level and mid falling tones, tend to have the Mal low tone.
Based on the results of our thorough investigation, we claim that language contact with
Khammueang, a variety of northern Thai dialect spoken in Nan, was the cause of tone birth in Mal. The high-
falling pitch seems to be typical in Mal native words, even in the non-tonal varieties. The tone shape [low-
rising] has been borrowed from Khammueang by Mal. It is worthwhile pointing out that a large number of
basic Khammueang words have either one of the two rising tones /low-rising/ (A1-2) and /mid-rising/ (A3-4).
The rising contour seems to play an important role for Mal bilingual speakers, i.e. it helps induce and
enhance the low-rising characteristic of the Mal low tone. The five stages of tonal evolution in Mal can be
postulated as follows:
*Stage 1: [ ] / [ ] (native words)
*Stage 2: [ ] / [ ] (native words and loanwords)
*Stage 3: [ ] / [ ] (native words) [ ] / [ ] and [ ] (loanwords)
*Stage 4: [ ] / [ ] and [ ] (native words and loanwords)
*Stage 5: /High/ [ ] and /Low/ [ ]
[ ]

At the present stage, Stage 5, the result of the acoustical measurement of the two tones suggests the
idea that, at the next stage (Stage 6) sometime in the future, Mal may have three tones: /High/, /Mid/ and
/Low/. Perhaps, the third tone could develop from the voicing of voiceless sonorants, e.g. /hm/ > /m/, /hl/ > /l/
and so on, as shown in Figure 8. More details can be found in Putthasatien (2007, 2009) and Intajamornrak
(2007, 2008, 2009a, 2009b, 2009c).

Figure 8 The average F0 values of vowels following voiceless and voiced bilabial nasals
in words having high tone
5. Conclusion
The results of our acoustic analysis confirm the claim of Sussman et al., that the LE method can help
identify the place of articulation of the three types of initial stop: bilabial, alveolar and velar.
The distinction between the average duration of short and long vowels with the ratio 2:1 is
statistically significant. This finding confirms the phonological analysis that Mal has a distinctive vowel
length. The smaller overall space of the short vowels shows their tendency to be more centralised than their
long counterparts. Moreover, the variation within the space of each short vowel is more dispersed. This
indicates the fact that the short vowels have more variants than the long vowels.
The birth of the /High/ and /Low/ tones was induced by language and cultural contact between Mal
and northern Thai. Five stages of tonal evolution can be postulated. The results of our study seem to suggest
that internal factors have had no role to play in Mal tonal evolution from the past to the present.
Vowels following voiceless initial sonorants have a higher F0 value or a higher pitch than those
following voiced ones, and the F0 difference is significant. This internal factor could be a cause of tonal
development in the future.
Acknowledgements
We would like to express our sincere gratitude to the Thailand Research Fund (TRF) and
Chulalongkorn University for funding our research project. Many thanks go to the local authorities of the
districts of Pua and Bokluea in Nan province and our Mal friends whose kind assistance and hospitality made
our field research possible.
References
Filbeck, David.1972. Tone in a dialect of Tin. Anthropological linguistics 14:111-118.
______.1976. Mal (Thin). In William A. Smalley (ed.), Phoneme and orthography: language planning in ten
minority languages of Thailand. Canberra: Australian National University.
______.1978. Tin: a historical study. Pacific linguistics, series B. no. 49. Canberra: Australian National
University.
______.1990. Mal phonology revisited. Mon-Khmer studies 18-19: 1232-1261.

Huadsiri, Yupaporn.2007. Identifying the place of articulation of initial stops with the locus equation method:
a case study of Hmong, Mien and Mal. Unpublished MA thesis, Department of Linguistics, Faculty
of Arts, Chulalongkorn University. (in Thai)
______.2009. An acoustic study of the place of articulation of initial stops in Hmong, Mien and Mal using
locus equation (LE) method. In Minegishi, Makoto, Kingkarn Thepkanjana, Wirote Aroonmanakun
and Mitsuaki Endo (eds.), Proceedings of the Chulalongkorn-Japan linguistics symposium. Tokyo:
CbLLE Tokyo University of Foreign Studies. pp. 173-182.
Intajamornrak, Chommanad. 2007. Tai loanwords in Mal: a minority language of Thailand. Paper presented
at The 9
th
Annual Southeast Asian Studies Graduate Student Conference, organized by the Southeast
Asian Program, Cornell University. 16 18 March, 2007. (To appear in Mon-Khmer studies 39)
______.2008. Fundamental frequencies of Tai Yuan tones spoken by Lua Mal speakers at Ta Luang village,
Pua district, Nan province. Paper presented at the 41
st
International Conference on Sino-Tibetan
Languages and Linguistics, organized by SOAS, University of London, England, 21 24 September,
2008. (To appear in Manusya, special issue 17)
______.2009a. Acoustical investigation of pitch patterns in Ta Luang Mal: implications for tonal evolution.
In Minegishi, Makoto, Kingkarn Thepkanjana, Wirote Aroonmanakun and Mitsuaki Endo (eds.),
Proceedings of the Chulalongkorn-Japan linguistics symposium. Tokyo: CbLLE Tokyo University
of Foreign Studies. pp. 203-212.
______.2009b. The acoustic characteristics of lexical pitches in four varieties of the Mal language of
Nanprovince: implications for tonal evolution. Paperpresented at the RGJ-Ph.D. Congress X,
organized by the Royal Golden Jubilee Ph.D. Program, Jomtien Beach Resort Hotel, Chonburi, 3 5
April, 2009.
______. 2009c. On becoming a tonal language of Mal-Pray spoken in Nan province.Unpublished Ph.D.
dissertation, Department of Linguistics, Faculty of Arts, Chulalongkorn University. (in Thai)
Jirananthanaporn, Supatra. 1993. Linguistic geography of minority groups in Nan province: a lexical study of
Pray and Mal dialects. Phitsanulok: Naresuan University. (in Thai with English abstract)
L-Thongkum, Theraphan. 2007. Final fricatives *-s and *-h in Proto-Tin lexicon. Manusya, special issue 13:
87-93.
L-Thongkum, Theraphan, Phinnarat Akharawatthanakun, Chommanad Intajamornrak, Kanitha Putthasatien,
Yupaporn Huadsiri and Supaporn Phalipat. 2007. Dictionary of the 15 languages spoken in Nan
province.Unpublished research report submitted to the Thailand Research Fund (TRF), Bangkok.
L-Thongkum, Theraphan and Chommanad Intajamornrak. 2009. Tonal evolution induced by language
contact: a case study of the Tin (Lua) language of Nan province, Northern Thailand. Mon-Khmer
studies 38: 57-68.
Minegishi, Makoto, Kingkarn Thepkanjana, Wirote Aroonmanakun and Mitsuaki Endo (eds.). 2009.
Proceedings of the Chulalongkorn-Japan linguistics symposium. Tokyo: CbLLE Tokyo University
of Foreign Studies.
Modarresi, Golnaz., Harvey M. Sussman, Bjorn Lindblom and Elizabeth Burlingame. 2005. Locus equation
encoding of stop place: revisiting the voicing/VOT issue. Journal of Phonetics 33: 101-113.
Phalipat, Supaporn. 2007. The relationship between duration and formant frequency of vowels: a case study
of Hmong, Mien and Mal. Unpublished MA thesis, Department of Linguistics, Faculty of Arts,
Chulalongkorn University. (in Thai)
______. 2009. A comparison of the acoustic characteristics of Hmong, Mien and Mal vowels. In Minegishi,
Makoto, Kingkarn Thepkanjana, Wirote Aroonmanakun and Mitsuaki Endo (eds.), Proceedings of
the Chulalongkorn-Japan linguistics symposium. Tokyo: CbLLE Tokyo University of Foreign
Studies. pp. 183-194.
Putthasatien, Kanitha. 2007. The behavior of fundamental frequencies influenced by voiceless and voiced
initial sonorants: a case study of Hmong, Mien and Mal. Unpublished MA thesis, Department of
Linguistics, Faculty of Arts, Chulalongkorn University. (in Thai)
______. 2009. The effect of voiceless and voiced initial sonorants on the fundamental frequencies (pitches)
of the following vowels in Hmong, Mien and Mal. In Minegishi, Makoto, Kingkarn Thepkanjana,
Wirote Aroonmanakun and Mitsuaki Endo (eds.), Proceedings of the Chulalongkorn-Japan
linguistics symposium. Tokyo: CbLLE Tokyo University of Foreign Studies. pp. 195-202.
Singnoi, Unchalee. 1988. A comparative study of Pray and Mal phonology. Unpublished MA thesis,
Institute of Language and Culture for Rural Development, Mahidol University.
Sussman, H.M., and Jadine Shore. 1996. Locus equations as phonetic descriptions of consonantal place of
articulation. Perception and Psychophysics 58: 936-946.

Appendix

McCormick, Patrick. 2011. Mon Converging Towards Thai Models. In Sophana Srichampa and Paul Sidwell (eds.)
81
Mon Converging Towards Thai Models:
Evidence from a Mon Historical Text
51

Patrick McCormick
University of Washington in Seattle
Language is central to the transmission of history, so that the actual form of language in a historical
text can reveal the contacts and cultural practices of its speakers. A language, in turn, is a mirror of social
circumstances and contact, of historical changes, of the history of social interactions and settlement, as well
as the vagaries of political affiliation and dominance. In the writing of Southeast Asian histories, the Mons
are considered to be one of the earliest groups to have arrived in the Mainland to have developed one of the
earliest, most advanced civilizations after having come into contact with Indic civilization and Buddhism.
This understanding has deep resonances outside the discipline of history, including in linguistics.
I take here examples of the language found in a Mon historical text to examine the intersection
between Mon history and the study of the Mon language to question the common assumption that similarities
between Mon and the neighboring languages of Thai and Burmese are always due to primary Mon influence.
The evidence of the text suggests that Mon-speaking community in 19
th
-century Siam, surrounded by
speakers of Thai, came to replicate Thai patterns of syntax and usage without necessarily borrowing many
word or word forms from Thai. Reflecting the social contingencies of a contact situation, these developments
are not surprising from a linguistic perspective, but from the perspective of Burmese or Thai history and
historiography, interpreting Mon influence as anything but primary, or to see the Mon language as being
worked upon by other languages, may be highly unexpected, troubling, or unacceptable.
The texts of the Rjvasa Kath
52
are a collection of historical and literary narratives about a
different period of Mon history. The longest and best-known component text, for which there are also Thai-
and Burmese-language retellings, is the narrative of Rjdhirj.
53
The narrative traces the rise of the
illustrious ancestors of Rjdhirj before turning to trace his career as a military hero, both against other Mon
polities and against the Burman court, spanning roughly the late 13
th
to the latter part of the 14
th
centuries
AD.
54
Rjvasa Kath is one of the names that a Siamese Mon monk, Nai Candakant, gave to a
collection of prose historical narratives. He compiled several independent historical texts and printed them in
book form in two volumes in 1912 and 1913 at Pk Lat, a Siamese Mon village now on the outskirts of

51
My thanks to Laurie Sears, Mary Callahan, Christoph Giebel, Charles Keyes, Jacques Leider, Mathias Jenny,
and Paul Sidwell for reading and commenting on earlier versions of this chapter. Thanks also to Christian Bauer,
Nicoletta Romeo, John Okell, and Justin Watkins for providing me feedback on presenting the data. Special thanks to my
Mon informants in Burma and Thailand, who I do not name here because of local sensitivities.
52
a a . . qo o(o ). _c. _ . . [Candakant, Nai. Rjvasa Kath (2
Vols). Pk Lat, Siam: 1911-1912, in Mon]
53
The relationship between the narrative in the three languages is not one of whole translation of one into the
other, although individual passages at times may be very similar. I avoid the term version, with the implication of an
original, and instead used the term retelling, which opens the possibility of transmission through oral recitation and
performance.
54
The Mon title is c o ooo _o o Galn Ktw Bdah Tnow Datow Smi u Has,
meaning, On the Origin of the Hasvat Succession of Kings. Unlike in the Burmese and Thai titles, the name
Rjdhirj is not present in the Mon, but I refer to it here as the Rjdhirj narrative for the sake of convenience.
82 Mon Converging Towards Thai Models

Bangkok.
55
The language of the Rjvasa Kath, particularly the narrative of Rjdhirj, appears highly
Siamified.
56
The narrative appears to have been either translated from Thai, or recorded as retold in a
Siamified dialect of Mon from what was then Siam. This evidence in the Mon text of the close contact
between the Thai and Mon languages has apparently escaped the notice of decades of Mon and Burma
scholars, including those Burma Mon
57
scholars living in Thailand who have become fluent and literate in
Thai.
I believe that the process of linguistic convergence, in which Siamese Mon has replicated Thai
linguistic models, is the most likely explanation for most of the unusual features of the Mon language of the
Rjvasa Kath. Involved here are languages in contact, which implies speakers in contact who are
embedded in a variety of social connections and interactions. There are not many sources on the Mons living
in Siam before the 20
th
century. Mons settled scattered throughout what is now Central Thailand. The usual
understanding
58
is that these Mons arrived from Burma in waves starting in the 17
th
century, although there is
mention of Mon-speaking communities at Ayutthaya in earlier centuries.
59
Mons cultivated rice and worked
as potters, while women engaged in trade and men could serve in the Siamese army in ethnically segregated
regiments. There were also Mon women at the Thai court. Rice cultivators were the first to assimilate to the
Siamese, whereas those living in more isolated communities, particularly if engaged in trade or an occupation
that the Siamese did not engage in, tended to maintain their language much longer.
60
There appears to have
been regular contact between Mons in both countries, particularly between religious institutions.We can thus
establish a social contact in which some of the features of the language of the Siamese Mon developed,
reflecting the contingencies of contact outside of Burma.
Linguistic Evidence for Convergence
61

I consider here some of the features of the language of the text of Rjdhirj as evidence of
extended contact with, and possible translation from, the Thai language. Of particular relevance are examples
of Mon prose, using native words, reshaped and reworked along the lines of Thai linguistic models. We
might say the text speaks Mon in Thai ways. Following this is a set of examples of this process of
convergence and replication.
Mon scholars, when asked about the unusual language of the Rjdhirj narrative, either deny that
there is anything unusual, or think of the language as archaic or full of errors. Many Burma Mons familiar
with the Mon language of Thailand think of it as preserving features now lost in Burma. While the Mon
dialects of Thailand do in fact preserve some vocabulary no longer used in daily speech in Burma dialects,
the modern spoken varieties display even greater convergence towards Thai models than does the language
of the Rjvasa Kath.

55
The printed text has been available in Siam and Burma for close to one hundred years and has been the focus
of scholarly attention.
56
I use the term Siamified because this process started before there was the modern state of Thailand.
57
That is, Mons from inside Burma, in opposition to the Siamese or Thai Mons, the Thai Rma.
58
See Robert Halliday, Immigrations of the Mons into Siam, in Michael Smithies, ed., The Mons:
CollectedArticles from the Journal of the Siam Society (Bangkok: Siam Society, 1986). Also Suphorn Ocharoen
, ( : , 2541) [Suphorn Ocharoen, The Mons in
Thailand, 1988]
59
See Victor Lieberman, Strange Parallels: Southeast Asia in Global Context, C. 800-1830, Volume 1:
Integration on the Mainland, vol. 1 (New York: Cambridge University Press 2003).
60
See Brian Foster, Commerce and Ethnic Differences: The Case of the Mons in Thailand. (Athens: Ohio
University Press, Center for International Studies, Southeast Asia Program, 1982). It is in some of these areas
connected to the surrounding areas by road only since the 1980s where Mon is best maintained in Thailand.
61
The first version of this chapter was conceived of without any knowledge of Bauers 1986 article on
structural borrowing in Thai Mon. Bauer has taken a very similar approach, examining examples of modern spoken Thai
Mon for evidence of contact with Thai. While I have taken the written language of the late 19
th
century as the basis of this
examination, Bauer has taken the modern spoken language. Nevertheless, our findings are substantially similar. See
Christian Bauer, Structural Borrowing in Mon: Towards Language Death? in Journal of Language and Culture vol. 6
no. 2 (Bangkok: Mahidol University) 1986.
Patrick McCormick 83

Yet at the same time, there are patterns and expressions in Siamese Mon that may be only partially
meaningful, or misleading, to speakers of Burma Mon, but which become clear when interpreted through the
lens of a Thai expression or usage. My evaluations of the naturalness and markedness of the language has
been strongly shaped by reading it together with my Mon language tutor, Nai Hawng Htaw.
62

Below I consider examples of reanalysis, the introduction of new patterns, grammaticalization, and
direct borrowings. Particular examples may represent more than one phenomenon. Reanalysis is a process
in which speakers reinterpret Mon words and word forms to mirror those of Thai. The results of are varied: at
times the reanalyzed form will exactly mirror the model language, while at others, hypercorrection or
incomplete learning may result in forms that do not match exactly the model language. Grammaticalization
is a very process in which words are reinterpreted to fulfill grammatical functions.
63
Many of the classifiers
widely found in languages of Mainland Southeast Asia have their origins in nouns, and some of the so-called
directionals or adpositions of Thai and Burmese have their origins in verbs and nouns. I also consider
examples of Pli being used in different ways between the languages of Burma and Thailand, and of a few
direct Thai loanwords. I further examine examples of what I have called translationese, language that is
still so close to the model or source language as to be unintelligible without reference to the model expression.
There is often a surprising congruity between Burma Mon and Burmese on the one hand, and a
disjuncture between Burma Mon and Thai on the other. This congruity in fact highlights a conceptual
stumbling block: many Mons think that Mon and Thai should be similar because of basic typological
similarities, such as word order, which is shared between Mon and Thai but not with Burmese, but in fact
Burmese and Burma Mon have come to share many patterns and even ordering of sentence elements.
Formal Possessive Marker
Unlike Burma Mon, Thai and Burmese can formally mark possession with the use of so-called
particles. While all three languages can use a strategy of juxtaposition, with either the possessor before the
possessed (in Burmese), or possessed before the possessor (in Thai and Mon), this latter is the only strategy
available in Burma Mon. In Thai, a common strategy is the grammaticalized use of the word k or thing,
which has lost its original meaning, but indicates posssession, as in:
Thai
64

knm k dk
snack POSS child
The above sentence, when rendered into Burma Mon, is:
Burma Mon
kwai kon
65

snack son/daughter

In the Rjdhirj narrative, we find a seemingly incongruous use of the word kr p, derived from
Sanskrit dravya, which means property, possessions; treasure. In fact, however, this appears to be a
replication and grammaticalization of the Thai use of k. Because of the possible reading of the Mon term

62
This is a pseudonym.
63
Following the definition of Payne, this is a process in which certain words take on a grammatical function,
thus losing their original meaning. See Thomas E. Payne, Describing Morphosyntax: A Guide for Field Linguists
(Cambridge, UK: Oxford University Press, 1997), p. 239 and p. 262.
64
According to Huffman, Khmer has a usage exactly paralleling the Thai example. See Huffman Thai and
Cambodiana Case of Syntactic Borrowing? Journal of American Oriental Society 93, no. 4 (1973): 489-509.
65
Burma Mons whom I have consulted reject the usage of this word as a possessive marker, although there is
evidence that Burma Mons who have settled on the Thai side of the border have already replicated this Thai pattern. See
Jenny 2010 (forthcoming). The Matichon dictionary of Siamese Mon does in fact list kr p as having the Thai
grammaticalized usage. See - - ( :
, 2548), p. 143. [Committee for the Preparation of the Mon-Thai Dictionary, Mon-Thai Dictionary, for Thai Mon,
(Bangkok: Matichon, 2005)]

as property, in some contexts the incongruity is not as apparent as in others. In all the examples below, the
normal Burma Mon was to express possession would be to drop the use of kr p.
p 301
66

q c o c _o g
Nr minl sm p .lp h kro .hta coi.ceh kr p
Narmilla with general-troop family elephant-horse THING
. _
ckao,
67
no i.ka c k tt kl po pra ckao
body, strenuously march go.out come direct side body
q q .
tm.ca tla r c tira t pln.
foot lord` Rjdhirj AGAIN
Narmilla, together with his troops, their families, and calvary, forced themselves on to march straight back
to the presence of Lord Rjdhirj.

In this above example, to a Burma Mon, it sounds as though Narmilla is also taking along his
possessions or treasures.
The next example is noteworthy for reasons other than just the use of kr p. This excerpt is from a
scene in which the speech of children or the mad is interpreted for portents of the future. Ba Noy, the
name of Rjdhirj when he was younger, has sent some of his followers to hear the news at Hasvat,
where some of his men have gone to a gate of the city to listen. We find words and phrases that are highly
reminiscent of Thai, but cannot necessarily be put back into Thai word-for-word. This example, as in many
that follow, exemplifies the linguistic slippage that can mark the language of Rjdhirj.
We find the presence above of an otherwise unknown Sanskrit word rjaastra, here r csat,
following the Thai pronunciation
68
rtcst. In Thai usage, this is a law promulgated by the
king in accordance with the principles of the Dharmaastra. Following the interpretational lens of Thai, at
first glance the phrase h m t h cno k implies, the person who becomes/will become great. A more likely
interpretation would be the common Thai expression, p pen yy, a common description of
someone who is a superior.
pg 194
_ o _ gg
cn a poa tra hs.ka k mo i pru kr p
when go in door T.K. GET hear sound THING
|q o
t
r ka m hm
child REL say

66
Unless otherwise indicated, page numbers are all from the Rjdhirj section of Phra Candakants
Rjvasa Kath.
67
The three lines provide script, phonetic transcription, and gloss. Words that are capitalized have been
grammaticalized. Abbreviations: DEIC deictic; DUR durative or progressive; FIN abbreviation for FINISH, acting as a
conjunction; FOC for focus particle; HON = honorific; INT intensifier; IRR and REAL for irrealis and realis; NEG
negative; REL relative; TOP topic; VOC = vocative; a period between two words reflects that they are equivalent to one
word in the other language; = indicates that one unit contains a combination of fused meaningful units. Numbers
followed by S or P indicate pronouns: 1P is the first person plural. In keeping with Burma Mon tradition, phonetic
renderings reflect reading, formal pronunciations rather than colloquial.
68
This word has come through Thai because of the treatment of the final consonants: if the Sanskrit had
descended through Mon, it would likely not result in a final t, but a final s, realized as h.

o qo o o
k h smoi cno k kl ht h r csat k t.cneh
SAY sming
69
great come establish rjaastra GET attack
_ . _
t .to kao h m t h cno k h m
accept with person REL be big. person REL
.
t h cno k m k cn h.
be be.big REL GET be.victorious
When they went to the Tasawng Kaing gate, they heard the sound of child saying,The great sming who
brings about the rjaastra will attack and repulse the superior man. The superior man will be victorious.

For purposes of comparison, the following is an example of the use of kr p following the normal
Burma Mon meaning of possession, treasure, in this case the latter, which reflects more closely the original
Sanskrit meaning. This example is taken from the oqooo Slapat Rjva Datow Smi
70
or
(Treatise) on the History of the Kingly Lineage, sometimes referred to in English as The History of
Pegu.
p 17 (Slapat Rjva Datow Smi)
o _ .
smoi in k a t ket m ta t kyaik pe te tuy,
King Indra GET see take
71
LIVE relic Buddha three those FIN
o gg q o q.
pa paoc .ska kr p r t h tm k pno t tw ra
do pj treasure jewel not know with extent limit FOC
Indra, catching sight of the three Buddha relics, made limitless offerings of treasures and jewels.
Quotation Particle
A feature of Burma Mon syntax that sets it apart both from both Thai and Burmese is that it lacks a
particle marking direct speech and other such complements. In Thai, the usual particle is w a
grammaticalized verb meaning say. The fact that w is used even with verbs meaning say, tell, indicates
that the original meaning has been bleached in the process of grammaticalization. In Burma Mon, a natural
way for speech to be quoted is to put the speech first, followed by a verb of saying or hearing. We also
commonly find examples of such verbs both preceding and following a quote.
We find in the text of Rjdhirj repeated examples of k h, a Literary Mon
72
verb meaning say,
tell, being used exactly following the Thai pattern of w. Consultation both with Burma Mon speakers and

69
Title of ambiguous meaning, from king through local leader or headman.
70
From Journal of the Burma Research Society vol. 31 no.1 (Rangoon: Burma Research Society, 1923), p. 15.
This text is often erroneously referred to as Slapat Rajava Datow Smi Ro, but the final word is an assertive focus
particle. Following tradition, I have given titles in Indic transliteration, not transcription.
71
According to Mathias Jenny (personal communication, March 2010), the usage of V+ket may be
grammaticalized, in a manner not parallelled in either Thai or Burmese, to mean to do something for ones own benefit,
in opposition to V+k, which means to do something for someone else.
72
As in many languages of Southeast Asia, there may be a sharp divide in the syntax, vocabulary, usage, and
even phonology between written and spoken styles of the language. The Burmese linguistic tradition
recognizes this distinction, with spoken language standing in contrast to literary language of writing and
formal speech. Burma Mons have varying degrees of control over the literary form of the language,
depending on education. Many areas of Literary Mon vocabulary have fallen out of use in the past century.

with contemporaneous texts from inside Burma has revealed that this usage appears to be unknown in spoken
Burma Mon. A complication to using the evidence of this expression is that k h is the second element in the
Literary Mon phrase hm k h, an elaborate expression meaning to say used in formal language.
73
Instances
of uses outside of this literary collocation may provide stronger evidence for the Mon replication of the Thai
model, as in:
p294
_ q q |
tla h r c tira t tm k h pnan hm cih
lord Rjdhirj know SAY army burman descend
q _ _.
r m prn.
surround town Pye.
Lord Rjdhirj knew that the Burmese army had come down and surrounded Prome.

The above example follows Thai word order, whereas in Burma Mon, the subordinate clause would
precede the main verb, at least in Literary Mon. We may note that this Mon construction of placing the
subordinate clause before the first is suspiciously similar to the natural word order of Burmese. A Burma
Mon rendering of the above might be:
q | q _ _
parao pnan hm cih r m prn
circumstances army burma descend surround town Prome
_ q q .
k h, tla. h r c tira t k tm ket.
TOP, lord Rjdhirj GET know TAKE

A final example from the Slapat Rjva Datow Smi displays a more common Burma Mon
strategy of enclosing speech with multiple, non-grammaticalized uses of k h, which is highly reminiscent of
the parataxis of oral texts. At times, it can be difficult to determine which instance of k h goes with which
instance of speech.
p 21 (Slapat Rjva Datow Smi)
o _ e q. q o
smot.pr moi tuy s.k h ra. t h ra saik k h
young
.woman hear FIN IRR.say FOC be FOC manner TOP

o c q q.
slah lmy m pdoa r t pe ra o t
1S free in life jewel three FOC. 1S accept
q o _ .
kill FOC young.woman say.
hct ra smot.pr k h.
Having heard that, the young woman said, In that case, I donate my life to the Three Jewels. I will be
killed, said the young woman.

73
An elaborate expression, found throughout the region, are made of compounds of words with the same or
similar meaning to give a feeling of weightiness to the occasion or subject matter.

Interrogative Strategies
There are many features of syntax that Mon shares with Burmese. Based on inscriptional evidence,
it appears that sentence-final question markers have existed in both languages (in unrelated forms) for
centuries. Burmese and Mon share a common interrogative pattern which stands in contrast to the Thai and
Khmer patterns.
74
Burmese and Mon usually indicate yes-no questions with an absolute question particle,
Burmese l and Mon ha.
75
Questions involving relative words, however, are marked with different particles,
usually l
in Burmese and rao in Mon.

76
Thai, in contrast, uses a variety of expressions to ask absolute
questions, including mi, often with an expectation of confirmation. Another strategy is r, often
followed by plw or not?, which has less an expectation of confirmation. The Thai-Khmer and the
Burmese-Burma Mon strategies are not directly equivalent and so do not always correlate with each other.
It is therefore striking when we find the seemingly incongruous use of Mon sentence-final ha in
mid-sentence or clause-initially. Sentences such as the following have led Burma Mon scholars to wonder
whether there was some kind of misprint or elision in the text:
p 201
co c q.
o cn
77
l lmy m kao mi ra
1s give.over PUT life with mother FOC
c o o a c ef ef
lmh w mi ket.na lmy m o ha ha
now this mother take life 1S INT INT
c .
pl h
78
lmy m o
free life 1S
(Ba Noy, having read a letter, is now speaking to his wetnurse) I have entrusted my life to you, my mother.
Now will you take or free it?

This sentence highlights the continued slippage between Thai Mon and Thai. The doubling of the ha
may occur because the speakers of the replicating language do not precisely match the model form, or the
form recorded is in flux before being established in place of the old usage. In the first Thai-Mon sentence, the
first instance of ha may be a continuation of the old interrogative pattern, while the second instance has taken
on the meaning of or following the Thai model. While sounding unnatural to the Burma Mon ear, the
above sentence does not actually sound acceptable when translated directly into Thai:
*
daw.n m aw ciwt cn r pl
y ciwt cn
Now mother take life 1S or free life 1S

The Thai word r by itself means or and can be used to indicate alternatives. The Burmese and
Burma Mon strategies to indicate alternatives, however, differs from the Thai. One natural strategy is to ask

74
See Huffman 1973.
75
Contemporary spoken Thai Mon often dispenses with the use of rao altogether, thus aligning the Thai Mon
usage with Thai, although such dropping is not unknown in Burma Mon.
76
Burmese has alternate forms for both absolute and relative questions, though these two are the most common.
77
This word is the result of a typical sound change in Thai Mon the Burma Mon equivalent is pyn, made up
of causative p + c n donate. Such sound changes sometimes have the effect of rendering common words into
something exotic-sounding to Burma Mons, who suspect that Thai Mon has preserved archaic words.
78
Despite the conjunct spelling, for the causative meaning, the initial must be realized as a sesquisyllable. Such
spellings are common enough and may reflect the uncertainty of the scribes when dealing with conjunct versus non-
conjunct forms. Central Thai script has no conjuncs, whereas most other Indic scripts do.

two parallel questions, each stating one alternative. The above second sentence, when rendered into Burma
Mon, would be:
c o c ef. c c ef.
lmh mi ket lmy m o ha, pl h lmy m o ha
now mother take life 1S INT, free life 1S INT

The last example sentence begins Thai and ends Burmese. The Thai honorific prefix phr
appearing in the title of Min Gaungs name, while at the end is the phrase h .sea pwa, highly reminiscent
of the Burmese phrase o o_ c m ho lo, a bad deed or something inappropriate. Of
interest here is the last clause, beginning Thai-fashion with Mon ha. Nai Hawng Htaw wanted to move this
ha to the end of the previous sentence, thinking that it might form a rhetorical question or might be a
typesetting error. At least as likely an interpretation is to view the phrasing in light of Thai syntax:
p393
_ _ a c a
ht h tla. h pra m a k k nm lo ik ne n k h
fact lord Phra Ma Ka GIVE exist letter bring this TOP
o _ o c. q a
ksp kr h n k lo n. ck h t h ekarat nm k
thought harm deep very. but be king exist with
o o . ef c o o .
t sta. ha cp lit.cnay h sea pwa
law loyalty INT think strategy not correct deed.
Bayin Min Gaung means great harm having this letter sent. But he must be loyal, being the King. Or is this
an ill strategy?
Politeness Strategies
Out of the three languages under consideration here, Thai has the most elaborated speech levels. The
language has relied heavily on foreign, particularly Khmer and Pli-Sanskrit vocabulary, taking many
loanwords from these languages to provide special sets of terms used in reference to royalty and monks. In
both Thai and Burmese, there are a variety of pronouns and particles available to indicate social distance
between speaker and hearer or speaker and referent, or to indicated attitudes of the speaker towards the
statement. Mon has the least elaborated honorific and pronoun system, with few distinctions made according
to status or gender. In the texts of recent centuries, we find a few simple pronoun and vocabulary differences
that index honorifics. Burma Mon has strategies to indicate politeness in making requests, suggestions, or
commands. As in Burmese, these usually take the form of sentence-final particles and verbs used as
softening strategies.
In Thai, a high-frequency strategy is the use of the verb k, literally ask for, request that has
been grammaticalized as a way to indicate politeness, often without a directly stated subject. A direct
translation of Thai k into Mon is at, which in Burma Mon is used with nouns and is not used as a
politeness strategy. Clauses introduced with at are found in Rjdhirj, and I observed that this usage was
again confusing for Burma Mons, striking in their seeming unnaturalness to the Burma Mon speaker. The
Burma Mon equivalent of this sentence would drop the at.
p 199
f c oq
at k p lp h sray coi ceh
request give officer soldier hero elephant horse

_ _ q _ .
k o.oik rm k p i
GIVE 1S.vassal so.that enough suitable LITTLE.
Give me enough soldiers and mounts to me (so as to be suitable).

The next sentence is another example of one which is redolent of Thai, but does not translate well
directly back into Thai, especially since, as in the previous example, the combination of at k suggests Thai
k hy, may (I) let (me). At the same, as with the above example under interrogative
strategies, a direct back-translation leads to something unacceptable to Thai speakers.
79

p 215
f _ o oq a
at k oik sk t ka kl.ro k kn na y
request give vassal IRR.get accept affair repay gratitude lord
c _ .
t t lmy m i
until end life LITTLE
Let me repay my gratitude to you, my Lord, until the end of my life.

To what extent was the above sentence acceptable to Thai Mons of the time? Did they find it
immediately interpretable? Burma Mons have difficulty in understanding it and reject it on the grounds of
vagueness and unnaturalness, with the result that such sentences are a frequent obstacle to reading the Mon-
language Rjdhirj. Here is a more typical Burma Mon example, from Wottu Mi Do Keh Htaw, The Story
of Golden-Nib Mi Dong from 19
th
-century Burma.
80
The following features paratactic repetition of the verb
request.
p 61 (Mi Do Keh Htaw)
_ o o o f
n no a m ptom mre .saik w p tist at tao
with way REL begin manner this, Bodhisattva request DUR
| o o c c o
ko po smoi m sm.k lnim.lnaik smot
permisson in lord father together mourn true
fg c q.
pnt pe o knat at tao lw ra
extent three day extent, request DUR always FOC
In this manner, beginning in this way, the Bodhisattva kept mournfully asking for permission [to go] from his
father for three days, he kept asking [sic].

A further politeness strategy, although not as widely found, involves the Mon word for help,
although sometimes the Mon rendering of the Thai word is found. Thai speakers can make requests for the
listener to do something on behalf of the speaker by placing cay help at the beginning of a sentence.
Such a strategy does not exist in either Mon or Burmese, and so the presence of verbs meaning help at the
beginning of sentences in the Rjdhirj narrative is conspicuous. There are two strategies for rendering
cay into Mon: translating it into its Mon equivalent r m ba , or using co , the closest phonetic rendering

79
A direct translation such as does not appear
acceptable, while may be more so.
80
co q . o _o. _ . o oa a . Nai Lokasri, ed., Wotthu Mi Do Keh Htaw
[The Story of Golden-Nib Mi Dong], (Mawlamyaing: Thuwunna Press, ND).

of the Thai into Mon. The idea of help appears weakened or removed from the Thai expression, so that
Siamese Mon speakers, when attempting to render the Thai into Mon may have preferred to introduce the
Thai expression. Burma Mon readers tend to interpret this usage literally, in the sense of give me help
and .
The first example uses both the Mon and the Thai expressions, and is closed with another Thai-like
usage of Mon km also, parallelling the positioning in Thai of day also in requests:
p 198
f g ig o _ .
mi pa co hm k i km
mother HELP HELP say GIVE LITTLE also
Mother, please tell him for me.

The second sentence is more elaborate and features both native (sentence-final i ) and borrowed
strategies:
p 209
c a ig o _ _ .
tlakn pyeh me tta co a hm k p no i
monk, throw mett HELP go say GIVE Ba Noy LITTLE.
Your reverence, please
81
go tell Ba Noy for me.
Grammaticalization
Having already considered the cases of turning thing into a possessive marker and say into a
quotation marker, I consider here two other cases of grammaticalization. The first is the use of the word
arrive to indicate a complement of verbs of thinking and feeling, and the other is the use of the word search
to indicate direction of motion or action towards humans. I have chosen these two for their relative frequency
in the text of Rjdhirj, but also because their interpretation may be deceptive to the Burma Mon reader.
Depending on the context, these usages appear either superfluous in a sentence, or seem to have a literal
meaning that is in contrast to their intended meaning.
Arrive
In Thai, the verb t can mean arrive but has a function of also marking direction towards the
endpoint of something. At the same time, it can also be used with more abstract concepts and means
something like about, concerning. The use of t is particularly common with the verbs like think of,
remember and say, talk about. In the following sentence, we have a clear example of the Thai usage together
with one of a a few direct Thai loanwords: the term kt means think, plan, consider, wonder, be of the
opinion. In Burma Mon, the various meanings of kt tend to be broken up into disparate expressions, some
borrowed from Burmese. Because kt is used quite frequently in Thai, it may be that speakers borrowed the
Thai word into Mon to fill a perceived gap in their own language, and is not some kind of incomplete
translation:

81
The collocation throw mett is used in both Thai ( p
mtta) and Mon to mean bless, but here

appears to be used in the sense of please, or perhaps do the favor of This usage is described for modern spoken
Thai Mon, where it is used as a polite expression, not having to do with monks:
c a _ _ .
at pyeh me tta klon klon n k o.oik i
request throw mett do work this GIVE 1S.vassal LITTLE.
Please do this work for me.
In Thai Mon, the pronoun o.oik is not used only when talking to monks but as a general polite 1S pronoun.
See , ( , 2526), p. 138. [La-or Paencaroen. Mon Grammatical
Systems. MA Thesis, Silapakorn University, 1983].

p 199
_ f g f g
k p no kt cp ckao tla me thao
GIVE Ba Noy think ARRIVE body Lord Me Thao
Let Ba Noy think of Lady Me Thao; let Ba Noy miss Lady Me Thao.

To the Burma Mon, the above sentence may sound something like, Let Ba Noy think arriving at
Tala Me Thao herself, although Burma Mons are not familiar with the Thai loanword. A Burma Mon
rendering of this sentence would be rather different, something on the order of the following, although the
Mon rendering cannot cover the same range of meanings as the Thai:
_ o
k p no k p h.sna tla me thao
GIVE Ba Noy
82
GET miss Lord Me Thao
Let Ba Noy miss Lady Me Thao.
Search
In Thai, another word used in a similar way is the h which has been grammaticalized to
indicate directionality towards humans. The verb literally means look for, but in many contexts, the idea of
actual searching is absent. This usage is often coupled with the verbs come and go.
83
In Burma Mon, as in
Burmese, there are ways of indicating directionality towards a human goal. In the following sentence, we
find a clear example of Thai Mon having replicated the Thai pattern, with the native Mon kla y, search, look
for being pressed into service. For a Burma Mon speaker, this usage is discordant because of the Burma Mon
speakers desire to interpret the meaning literally:
p 210
_ f q.
k p no tt kl kla y o ra
GIVE Ba Noy exit come search 1S FOC
Let Ba Noy come to me, Have Ba Noy come to me.

In Burma Mon, this sentence could be rendered:

_ eg q.
k p no tt kl cr o ra
GIVE Ba Noy exit come vicinity 1S FOC
Let Ba Noy come to me, Have Ba Noy come to me.

Here is an example of kla y from Wotthu Mi Do Keh Htaw, which shows a more typically Burma
Mon usage of the word:

p 115 (Mi Do Keh Htaw)
c o .
lp a kl uca n smoi tuy,
enter go orchard garden king FIN

82
When rendered in Mon, Thai names tend to be rendered following the Indic values of their letters. Modern
Mon has neutralized the distinction between graphic OY and AY, both now o, so that spellings can be
interchangeable. I suspect this is a representation of Thai n
y, small or junior.
83
There is some similarity between the Thai and English in such sentences as, Come find me in the library
tomorrow, where there is not an implication of an actual search.

f o
o q o _ o.
kla y h i st.cu r h ct krk pnh.
search.for pluck eat fruit root bone mango jackfruit
They went into the garden of the king, searched for, plucked and ate the measly mangoes and jackfruits.
Changing see into think.
A final set of expressions, admittedly chosen out of a very wide range of choices, have to do with a
Thai usage covering a broad range of meanings. These include ideas of thinking, planning, considering, and
agreeing, all based on on the Thai verb hn, or see in English. The usage takes on a variety of meanings
depending on the complementizer it is coupled with. These include hn see; plan; consider; hn w be of
the opinion; think; and hn day agree, be of an accord.

The narrative of Rjdhirj features examples of the direct equivalents of each of these expressions,
at most only marginally intelligible in Burma Mon. The Burma Mon equivalent of Thai hn is a t, which can
only mean see and does not match the semantic range of the Thai. The expression hn w, literally see say,
the second element is the grammaticalized quotation particle discussed above. The Mon rendering of a t k h
appears meaningless to Burma Mon speakers. The reader frequently meets these usages throughout
Rjdhirj, but because of their idiomatic, extended meanings, they posed significant hurdles towards
comprehension for the Burma Mon speakers that I consulted. This was especially the case with the Mon a t
km, rendering hn duy, which sounds like see also. The expression see say is one of the most frequent in
the text, in large sections occurring on nearly every page.
p 202
dfg co q.
a t k h t h i.t l kla ra
see SAY HIT endure set-down before FOC
(I) think we will have to endure it for the time being.

In Burma Mon, this might be rendered as:
co q. .
t h i.t l kla ra, o te

HIT endure set-down before FOC, 1S think

We may note that the above Burma Mon sentence not only employs a Mon rendering of the
Burmese t, English think, have an opinion but even replicates Burmese word order by placing the
quotation before the verb, as is typical of SOV word-order languages like Burmese.
Miscellaneous Syntactic Examples
The following are several less common examples of expressions that follow a Thai model, including
changes in word order and calquing. For example, Thai and Siamese Mon agree in the ordering of
collocations, with the head following the verb. Burma Mon and Burmese, however, have the head before the
verb. The order verb-head may be prototypical for SVO languages like Thai and Mon, so it is possible that
Burma Mon has changed under pressure from Burmese, or that Siamese Mon has preserved an older feature,
or redeveloped it through contact with Thai. In the narrative, we find apparently reversed expressions:

t ct
use.up mind
be fed up, be uninterested


This is rendered in Burma Mon as ct t.
84
The apparent reversal is frequently found in collocations
involving the metaphor of mind, the figurative seat of emotions in Mon, Thai, and Burmese. Here we have
a Thai expression calqued into Mon:
p 204
c fg f g | q.
lpa cp on ct po p .pnan poy ra
do.not think few mind in forces 3P FOC
Do not feel insignificant because of our forces.

To a Burma Mon, this sentence sounds something like, Do not feel sad because of our forces, an
interpretation that does not fit easily with the context. Rather, the sentence appears to be a rendering of the
Thai collocation n
y cay literally, few mind, meaning rather feel small, feel inferior This
interpretation fits better with the context. A Burma Mon rendering might be the following, which following
common Literary Mon usage, reverses the two clauses in a Burmese-like manner:
| c f g dfg q.
po p .pnan poy k h, lpa ct ot ra
in forces 3P TOP, do.not mind small FOC
Calque of the Thai Collocation k tt
The Thai expression to express ask for forgiveness, also used as a polite expression meaning Im
sorry, is made up the collocation of k, request as discussed above, and the Sanskrit tt,
meaning crime, offense; sin, fault. Burma Mon speakers use strategies meaning absolution and forgiveness
which are not exactly equivalent to the Thai. There is no Burma Mon expression directly equivalent to the
Thai, so that the following sentence sounds to a Burma Mon speaker as though a sin is being requested. It
makes sense only when interpreted through a Thai lens:
p 391
o o .
at th smoi p .na y t p
request fault smi leader army
....(they went in and) asked forgiveness of the leader of the army
Thai Pli
The central position of Pli both to religion and intellectual thought in Theravada Buddhist
Southeast Asia is well-known, yet the specifics between how languages of the region make use of Pli
vocabulary varies in detail. Just as the usage of Latin and Greek words varies between the languages of
Europe, so do the meaning and usage of individual Pli and Sanskrit words vary between Thai, Burmese, and
Mon. Compared with Thai and Khmer, there are fewer Sanskrit loanwords in the languages of Burma,
reflecting the narrower role of Sanskrit learning compared to Pli, at least in recent centuries. Throughout the
texts of the Rjvasa Kath, there are a number of Pli and Sanskrit words commonly used in Thai but
unknown or used with a different meaning in Burma Mon and Burmese.
85
With few exceptions, these words
are rendered in a Pli or Palicized form in Mon. Below is a short list of some of the more notable examples. I
have given the meaning of the Sanskritic vocabulary according to the Thai usage. The Mon pronunciations
are often educated guesses, as the words would be pronounced in Literary Mon.

84
This particular expression has no direct Thai equivalent. It is, however, exactly parallel to the Burmese
expression of the same meaning, se ko.
85
As far as I have learned, the usage of Pli and Sanskrit does not vary much between Burmese and Mon, with
the exception of some old loanwords into Mon.

Mon
86
Pli/Sanskrit Thai Usage Comment
a o
tec np w
tejnubhva
kings power,
influence
No such specialized usage appears to exist.
o
pte sa
pradea
country
The Pli word means variety in Burma. In Thai Mon,
this term replaces _ city in the collocation
_ kwan country, creating o
kwan pte sa
o
pyaocn
prayojana
(Skt)

use, benefit
The Pli a kn from Pli gua is used with similar
meaning.

pna
pah
problem
Mon can use a Sanskritic form, _
prtt or _o prasna, prana.
o
wi ti
vidh
way, method;
customary
usage
The Thai usage covers two meanings not connected in
Mon. For method, no or o n o is used. For
custom, usage, kak kna can be used.
oo
smoya
samaya
age, period
ket derived from Pli khetta is used.
oo|a |
swannkot
svargata
(Skt)

go to heaven,
ie, die
Instead of Plifying the spelling to something like -
*saggagata, the Thai pronunciation has been rendered
in Mon
ooa
saowne
sauvanya
(Skt)

queens
command
No such specialized usage appears to exist.
o o
sneh
sneha (Skt)
charm, appeal
AMon word of this shape means landing place. There
is no exact equivalent to the Thai term.
a _
na ta
anuta
(give)
permission
In Burma, this rare word means nonentity. In Thai, it
is used as a verb, for which Mon uses | k
ko.
87

c
itti bl
iddhibala
power, force
Mon uses various words inherited from Pli,
including a ana, aoc , or o isa
conn. Pli sa; or native o w .
o
asoa
aya (Skt)
dwell,live,
reside
A variety of native expressions, m , literary
tao, or pa htn or pa htao, can all be
used.
Table 1: Mon Words Following the Thai Use of Pli-Sanskrit
Thai Loanwords
The introduction of loanwords actual material from one language into another is not necessarily
as common as might be expected in contact situations, especially where there is long-term bilingualism.
Interestingly, other than the Pli words considered above, there appear to be few direct loanwords from Thai.
Some of the notable example of those that do occur include e meaning (one)self, which is borrowed
into Mon as , probably oi. We find a curious usage of the word o
o, which in Mon means bean, pea,

but from the context appears to be the Thai word b, rice-gall midge.

86
Some of the Thai Mon pronunciations are conjectural.
87
The second element is derived from an earlier pronunciation of Burmese, now kw but once
something like kwa .

p 219
o q . _
smoi m r t h ca kon mat tup pma
Smi Father R be only child minister be.like example like
o ih o o _ o o o
sne o. plah sne p pra k smoi sp
wing boa extend wing fly compete with king garuda
c o o.
l o kl m asmit km mk h k h kl ma n km ha.
sail cross sea also if TOP, cross can also INT.
Smi Ma R is only the son of a minister, like the wings of the boa. If he were to extend his wings and fly in
competition with Garua to cross the sea, would he really be able to?

The sentence is meaningful only when we take the word boa with its Thai meaning of a small insect.
The following sentence illustrates an isolated instance of the use of the word m ai, which in Mon means
corrupted, rotten, rendering the sentence unintelligible:
p 382
q oo o o |
m ra y ky.cwa mo i saik k h tuy poy c k pnan
Ma Ry Kayawchwa hear way that FIN 3P march army
c a
kl lo n n k h m ay k h pc coi kao
come occasion this DEIC intend SAY cause.fight elephant with
o _ _ .
smoi m n i
smi Mon LITTLE.

Taking the word at its Thai meaning, my in English, mean, intend, have the meaning, expect,
the whole sentence makes sense: Min Ye Kyawzwa heard what [his astrologer] had to say [and said,] Our
coming this time was meant to be to engage with the Mon sming in elephant battle.
Translation or Retelling?
As some of the examples already considered have suggested, many sentences in the Mon-language
Rjdhirj do not make for good Literary Mon as understood in Burma. Yet these sentences do not form
smooth Thai when translated word-for-word. An assumption of many scholars when dealing with literrary or
historical narratives found in more than one language tradition is that one must be a translation from the other.
A survey of the narrative in Rjdhirj in its Burmese, Mon, and Thai retellings does not suggest a complete
direct translation between any of them, or at least, between any of the texts extant today. The overall
grammatical and syntactic coherence of the Thai Mon of the Rjdhirj narrative, especially when viewed
through the lens of convergence, suggests that much of it has been recorded in a Thai Mon dialect reflecting a
noticeable degree of convergence towards Thai models.
88
Nevertheless, there are sentences and expressions

88
The modern Thai Mon dialects display an even higher degree of convergence towards Thai. In addition to
Bauers work cited above, see also for example , , ( :
, 2528 [1985]) [Champa Yeuangcharoen & Chamlong
Saraphatneuk, Learning Mon (Bangkok: Southeast Asian Language Department, Faculty of Antiquities, Silapakorn
University, 1985)]; and , ( , 2539
[1996]) [Phramaha Jaroon Chork Samud, Study of the Use of Words in Mon, (Bangkok: Silapakorn University, 1996)].

in the Rjdhirj that appear to be unintelligible. These could be examples of direct, though incomplete,
translation; examples of grammatical developments in Thai Mon that have neither a Burma Mon analogue
nor a Thai model; or examples of expressions that were intelligible at a certain time in a particular context,
but are not any longer.
The first example is quite startling, particularly because there is a similar, though not exact, sentence
in the main, most common Thai-language retelling of Rjdhirj. The meaning of the Thai sentence is, It is
up to you, my Lord, although the Thai Mon sentence is wholly unintelligible. The heart of the expression is
the formal Thai st t
, meaning be up to (ones wish), depend on (someones wish). The literal

meaning of the expression is ending from, from which it has taken on its extended meaning. The first element,
st, has been used in the Mon sentence. Perhaps the expression as it stands had become common in Siamese
Mon at the time. Alternately, the Mon speakers who recorded this text may have been unfamiliar with the
Thai and were unable to render an adequate translation. In this passage, one person is telling another that he
will work on her behalf, to which she replies:
p 198
o q.
sut ca tla o ra
sut only lord 1S FOC

The Mon sentence, which appears in any case incomplete, is wholly unintelligible without reference
to the Thai. The native Mon sut means silk or stra, suggesting something like Only silk as much as My
Lord. The following Thai sentence, however, makes the apparently intended meaning clear:
p 92 Thai-language Rjdhirj of Hon
89

sut t
pr.cw p kt t
t
up
.to Lord older.sibling think EXHORTIVE

As you like; Its up to you, my Lord.

A page later is another obscure passage, one that does not appear to have a direct Thai equivalent in
Hons printing of Rjdhirj in Thai:
p 199
| _ q q.
kt a uppay h.k h rao t ra
think go stratagem who even use.up FOC

Parsing the sentence to render h.k h as who, the usual meaning of the collocation, does little to
elucide the meaning when put into Thai, which is ungrammatical. If we reparse the expression h k h,
which could then mean that person, another meaning is possible:

kt c pay bay kw nn my mt lw
think IRR go stratagem 3S that, not use.up PERF

This second Thai sentence then means something like, There is no end to his thinking of stratagems,
which fits the context. In any case, the sentence is not wholly acceptable either when read as it is in Mon or
when it is recast in Thai.

89
See , () ( :
, 2423 [1880], [Chao Phraya Phrakhlang (Hon), Rjdhirj of Chao Phraya
Phrakhlang (Hon), (Bangkok: Silapabannakhan Press, 2001 reprint).

Conclusion: Intersections Between Historiography and Linguistics
Local Mon scholars must often work to make the Rjdhirj narrative meaningful. This work is
necessary in part because the reality of modern-day Burma is far removed from the context in which the texts
were created. More widely, ideas about history have influenced lines of linguistic inquiry. A persistent theme
is that the Mons and speakers of Austro-Asiatic languages are originary because they were among the first
peoples to appear in the historical record. Many scholars take this understanding of the Mons as the
underlying explanation for similarities between Mon and other languages. The logic of this argument may be
hard to escape if the Mon language of the Mon Rjdhirj has been influenced by Thai, then Thai itself
was influenced by Mon. The risk is that such an assumption may render other interpretations invisible. Much
work has yet to be done to determine the extents and limits of Mon-Khmer primacy. Directionality and the
contingincies in the convergence between Mon and Burmese for at least the past millennium remains to be
carefully studied. A survey of Mon literature since the 17
th
century suggests an increasing harmonization
from Mon towards Burmese. Recognizing the contingencies of specific situations may go far to render visible
the linguistic and social context that the language of the Rjdhirj narrative revealsa widespread
bilingualism among Mon speakers surrounded by Thai speakers for several centuries, resulting in the
convergence of Siamese Mon towards Thai models.
REFERENCES
Aikhenvald, Alexandra Y. Grammars in Contact: A Cross-Linguistic Perspective, in A.Y. Aikhenvald &
R.M.W. Dixon, eds. Grammars in Contact: A Cross-Linguistic Typology. Oxford University Press,
2007: pp. 1-66.
. Mechanisms of Change in Areal Diffusion: New Morphology and Language Contact. Journal of
Linguistics vol. 39, (2003): pp. 1-29.
Aikhenvald, A.Y. & R.M.W. Dixon. Dependencies between Grammatical Systems. Language vol. 74, no.
1 (1998): pp. 56-80.
. Introduction, in A.Y. Aikhenvald & R.M.W. Dixon, eds. Areal Diffusion and Genetic Inheritance:
Problems in Comparative Linguistics. Oxford University Press, 2001: pp. 1-26.
Bauer, Christian. Morphology and Syntax of Spoken Mon. PhD Dissertation, School of Oriental and
African Studies, University of London, 1982.
. Structural Borrowing in Mon. Towards Language Death? Journal of Language and Culture 6, no.
2 (1986).
Halliday, Robert. A Mon-English Dictionary. Rangoon: Mon Cultural Section, Ministry of Union Culture,
Government of the Union of Burma, 1955.
Heine, Bernd & Tania Kuteva. Language Contact and Grammatical Change. Cambridge University Press,
2005.
Huffman, Franklin. Burmese Mon, Thai Mon, and Nyah Kur: A Synchronic Comparison, Mon-Khmer
Studies vol. 16-17, (1990): pp. 31-84.
. Thai and Cambodian - a Case of Syntactic Borrowing? Journal of American Oriental Society vol.
93, no. 4 (1973): pp. 489-509.
Jenny, Mathias. The Verb System of Mon. Arbeiten Des Seminars Fr Allgemeine Sprachwissenschaft.
Zrich: Universitt Zrich, 2005.
Matras, Yaron. Language Contact, Cambridge Textbooks in Linguistics. Cambridge: Cambridge University
Press, 2009.
McCormick, Patrick A. Mon Histories: Between Translation and Retelling. PhD Dissertation, History
Deparment, University of Washington, 2010.
Shorto, Harry L. A Dictionary of Modern Spoken Mon. London: Oxford University Press, 1962.

Miller, Michelle M. and Timothy M. Miller. 2011. A Study of Language Use and Literacy Practices to Inform Local
Language Literature Development among Khmu in Thailand. In Srichampa and Sidwell (eds.) Austroasiatic Studies:
papers from ICAAL4. Mon-Khmer Studies Journal Special Issue No. 2. Dallas, SIL International; Salaya, Mahidol
University; Canberra, Pacific Linguistics. pp.98-111.
98
A Study of Language Use and Literacy Practices to Inform Local Language
Literature Development among Khmu in Thailand
Michelle M. Miller Timothy M. Miller
Payap University Linguistics Institute SIL International
SIL International
Introduction to language development in Khmu
The Khmu are a major ethnolinguistic group of northern Southeast Asia. The largest proportion of
the estimated 700,000 Khmu resides in Laos, and smaller communities are found in Vietnam, Thailand, and
China (Gordon 2005; National statistics center of the LAO PDR 2007). Khmu is a Mon-Khmer language and
encompasses a complex web of ethnic and dialect groupings (Chazee 1999; Suwilai 1987).
Khmu was first documented in the late 19
th
century by French scholars in Luang Prabang, Laos
(Preisig 1994). Systematic linguistic analysis, documentation, and efforts toward language development
began among Khmu in Laos in the mid-1950s with the work of William Smalley (Smalley 1961), and have
been continued by numerous other scholars (Lindell 1974; Lindell et al 1981; Svantesson et al 1994; Suwilai
1987). A small body of Khmu literature has been produced using both Lao- and Roman- script
orthographies
90
that are based on an Eastern Khmu dialect.
91
(Preisig 1990; Suksavang et al 1994;
Suksavang and Preisig 1998).
During the past 30 years there has also been extensive linguistic research conducted among Khmu
speakers in Thailand (Suwilai 1987, 1998, 2001; Cholthissa 1988) and there have been experimental efforts
to apply this research toward the development orthographies (Suwilai 1990; Supatra 1988; Cooper 1998,
1999), but currently no writing system has community endorsement nor is in wide use among Khmu
communities in Thailand for the development of Khmu language literature.
Research purpose: Informing Khmu language development in Thailand
In 2008, we (Timothy and Michelle Miller) were in contact with members of Khmu communities in
Thailand who expressed interest in developing an orthography representative of the dialects spoken in
Thailand. There already seemed to be adequate linguistic research upon which to base orthography
development, but there was no current documentation of sociolinguistic or literacy data to inform both the
process and products of a language development effort.
Consequently, in July and August of 2008, we undertook a study of a cluster of Khmu communities
in Nan province, Thailand to gather data in four areas:

90
An orthography is a standardized system for writing a particular language. The notion includes a
prescribed system of spelling and punctuation (Benson and Kosonen 2010:2 cf Crystal 1999:244).
91
Also referred to as Southern, particularly in earlier literature. This refers to the Khmu Uu (or Khmu Cueang)
and related groups who use the negative particle am and are found in Luang Phrabang, Phongsaly, Xiengkhouang,
Houaphan, Vientiane, Bolikhamxay and Sayabouly.
Michelle M. Miller, Timothy M. Miller 99

1. Language and culture maintenance and shift
To inform strategies for safeguarding Khmu language and culture
2. Standard Thai literacy skills and practices
To inform instructional approaches and the design of relevant materials
3. Attitudes toward language development
To inform community participation in a potential language development process
4. Perceptions of educators working in Khmu communities regarding the adequacy of Khmu
childrens Thai language skills for successful classroom learning
To inform plans for a potential local language instructional approach in the school setting

We acknowledge that the scope of our study was limited to only five of the estimated 30 Khmu
villages in Thailand, but from informal contact we have had with other Khmu villages we believe that what
we learned is representative of at least some of the other communities, and we hope that this study will
stimulate a broadened research and documentation effort.
In the remainder of this paper we will briefly introduce the research setting, summarize key findings
from our study, and present several recommendations.
The research setting
Thailand is home to an estimated 10,000 to 30,000 ethnic Khmu (Gordon 2005; Aguettant 1996).
Most of the Khmu villages are situated along the border with Laos in Chiang Rai and Nan provinces, though
there are small pockets of Khmu in Lampang and Uthaithani provinces (Suwilai 2002b).
The Khmu village settlements within Thailand are fairly scattered and have been established for
varying lengths of time. Many villages in the Lao border regions trace their histories to migrations from
Laos within the last 50 years, while other villages have been established for much longer periods of time
(Suwilai 2002; Supatra 1988).
We chose five villages in Chon Daen sub-district of Nan province as the site of this research because
this sub-district has the highest concentration of Khmu speakers in Thailand. Khmu in these villages identify
themselves as part of the Tmooy Thruel clan and speakers of a common dialect (Miller & Miller 2009). Their
dialect may be classified as part of the Western Khmu dialect grouping which includes Khmu referred to as
Khmu Rook in Laos (Svantesson & House 2006; Chazee 1999).

Village name Thai name Local Khmu name (IPA) No. of households Population
Nam Paan . /om pa:n/ 65 450
Huay Klaep . /om klp/ 71 378
Nam Lu . /om l/ 74 453
Huay Moi . /om mj/ 24 244
Ban Mai Chaidaen . /om sn/ 39 209
Total 273 1734
Table 1: Name and population of Khmu villages in Chon Daen sub-district, 2008
(Songkhwae District Office registrar, July 2008).
Research Methods
Data for this research project was gathered during a 6-week period in 2008 using questionnaires,
interviews and personal observation. Nearly 70 respondents participated in this study, and informed consent
was given by all respondents.
100 Language Use and Literacy among Khmu

Research Findings
1. Language maintenance or shift
It has been well documented that extensive language shift to a regional language or to Standard Thai
(the de facto national and official language) is taking place among many of the ethnic minority language
communities in Thailand (ILCRD 2008; Vail 2006; Tehan & Nahas 2009). Fourteen of the nations 74
documented languages are currently deemed severely endangered by the Institute of Language and Culture
for Rural Development (ILCRD), Mahidol University, Thailand (ILCRD 2008) and many others can likely
be classified as unsafe to use the terminology of Brenzinger et al (2003). Previous to this study the
language use within Khmu communities in Thailand had not been documented for 20 years.
There are a several analytical frameworks that have been used to identify particular points of
vulnerability with respect to language vitality and to prioritize strategies for language safeguarding or
revitalization, which include:
Joshua Fishmans Graded Intergenerational Disruption Scale (Fishman 1991)
UNESCOs proposed nine factors on language vitality and endangerment (Brenzinger, Matthias et
al 2003)
Landweers eight Ethnolinguistic Vitality Factors (Landweer 2000)

In our analysis we will present five key factors that seem particularly salient in understanding the
situation of the Khmu.
92
The indicators we will discuss come primarily from Fishman (indicator 1) and
Landweer (indicators 2-5 below). We are using the Landweer (2000) framework because it brings the issue
of economic security into particular focus.
1) Intergenerational transmission of the mother tongue
2) Population and group dynamics
3) Domains of language use
4) Access to a stable and acceptable income base within the locale
5) Social outlook regarding and within the speech community
1.1. Intergenerational transmission of the mother tongue
Khmu children are currently learning Khmu in the home, and it is the primary language of
communication between all three generations within most households. All of the respondents to our survey
(36 of 36) reported that they can speak Khmu, Muang (the Nan-Phrae province variety of Kammuang or
Northern Thai) and Standard Thai (ST). Ninety-four percent of respondents reported that Khmu is their
strongest language.
The data below provide further documentation of how language is being used in the home:
Between spouses: Khmu was the language spoken between husband and wife in over 80% of the
homes surveyed. In the remaining 20% respondents report that Muang or ST is used because one of
the marriage partners is not Khmu.
With parents: All respondents reported that they speak Khmu with their parents.
Parents with their children: 89% of respondents reported speaking Khmu with their children.
With siblings: 97% reported speaking Khmu with brothers and sisters.
First language of children in the village: 89% reported that Khmu is the first language children
speak and 6% reported that Khmu is learned simultaneously with the mother tongue of a non-Khmu
parent. Thus, 95% said that Khmu is the first language of the children.
The teachers at village Early Childhood Development Centers reported that the children cannot
speak any Standard Thai or speak just a few words, such as hello or eat rice when first

92
Note that a more complete analysis is presented in Miller & Miller 2009.

entering these state-sponsored daycare programs. This was further evidence that Khmu is indeed the
language spoken at home.

According to Joshua Fishman (1991) transmission of the language in the home is the threshold
level for language maintenance, the level at which small languages continue to survive and even thrive (cf.
Lewis 1996:8; Fishman 1991:92). We thus assess this finding as a clearly positive indicator of Khmu
language maintenance.
1.2. Population and group dynamics
According to Songkhwae district records, the absolute populations of the villages that participated in
this study have been stable and have even shown growth during the past 20 years (Miller & Miller 2009;
Supatra 1988). There are only a few non-ethnic Khmu living in the villages because of marriage to a Khmu
spouse.
Some significant changes are taking place, however, in the group dynamics of the Chon Daen sub-
district villages. There has been a dramatic increase in the number of young people, particularly women,
who are seeking work outside the village after completing their education. (Most young people typically
complete secondary grade 3or mathayom 3). Of female respondents 35 and older, none had ever worked
outside the village, but 78% of women ages 15 to 34 had worked in an urban center for a period of several
months or longer, some on multiple occasions. It should be noted that our sample population included only
people currently living in the village.
Male respondents reported that employment outside the village was not a recent development, as all
male respondents in both older and younger groups had worked outside the village on one or more occasions.
People most commonly reported finding employment as factory laborers, gas station attendants,
construction laborers, and drivers. Forty percent of respondents reported working outside the village on
multiple occasions for time periods ranging from six months to two years. Some reported seeking work on a
seasonal basis following the agricultural cycle, while others sought more permanent work and come home
occasionally during holidays. It should be added that not everyone who goes to the city to work finds the
experience to be financially profitable. Several respondents noted that women generally have more success
saving money from their urban work experience than the men who tend to spend more money on alcohol.
The increase in the number of Khmu youth seeking work in urban areas has led to changes in
marriage patterns. Village leaders observed that very few young people are currently marrying other Khmu.
Other respondents estimates of the number of Khmu youth marrying non-Khmu ranged between 50 and 90
percent.
The question of what effect this will have on the language of the children appears to depend on
where the children are raised. We do not have clear data indicating how many young people who marry
outsiders return to the village, but our data describes several scenarios. One respondent noted that sometimes
marriages fail and the Khmu spouse, generally the wife, will return to the village to raise the child. We met a
number of young mothers in this situation. Several respondents explained that sometimes the children of
mixed marriages are left in the care of grandparents in the village while the parents work in the city. In these
cases the children speak Khmu as their mother tongue. One respondent noted that a number of young women
had married young men from northeastern Thailand and had moved to the Khmu village because no suitable
farmland was available in the spouses home area.
When we asked both Khmu village leaders and individual respondents about community attitudes
toward marriage with non-Khmu, responses were similar. Among the survey respondents, none expressed an
unfavorable opinion toward youth marrying non-Khmu, with the exception of one respondent who reported a
taboo forbidding Khmu to marry a person of Htin ethnicity. Most respondents expressed the opinion that, It
depends on the couple to decide who they will marry, and many expressed that the key issue should be love
between the man and woman, not ethnicity. Interestingly, two respondents said they hoped to marry
someone from Bangkok in order to raise their social status.

There is evidence in our data that this current ambivalence toward marriage with non-Khmu is a
more recent trend, and that in the past marrying a fellow Khmu was an important value. One respondent said
that in the past it was an important cultural value to marry a Khmu, but that now it does not matter. Kamnan
(sub-district chief) Min explained his perception of how attitudes and practices toward marriage have
changed, and expressed concerns that parents have related to intermarriage:
Now only a very small percentage of Khmu marry fellow Khmu from our villages.
In the past few years only a small percentage of Khmu young people has married within
the Khmu tribe. In the past, people might arrange a marriage between neighbors or
distant relatives from the time their children were young. In doing so parents could ensure
that the person their child was marrying was diligent and from a good family. Now they
have no way of helping ensure that their children's marriage is stable. They do not know
whether the person might have AIDS or something else. When some in-laws (outsiders
married to village children) come to visit the village, their in-laws will not speak to them.
There is a feeling that they have stolen their child away to live elsewhere, and there are
bad feelings about this in some cases (Kamnan Min).

Regarding connectedness with the broader population of Khmu, the residents in Chon Daen sub-
district villages reported only minimal contact with other Khmu communities in Thailand outside Nan
province. For most people, contact with Khmu from outside the local district was reportedly rare. Chon Daen
residents do have fairly regular contact with Khmu from Laos, though, at a weekly border market.
Landweers (2000:12) language vitality scale regarding population and group dynamics brings into
focus the influence of the migration of outsiders on the ethnolinguistic vitality of local language communities.
However, in this Khmu context, while there is some in-migration of outsiders, a more relevant appraisal of
changes in population and group dynamics is linked to the increasing number of young people, particularly
women, moving out of the village to seek employment. There are currently only a few non-Khmu who have
migrated into the Khmu villages, and many of these have learned to speak Khmu.
Given this observation we have modified Landweers scale as follows. Note that a score of (4)
indicates high ethnolinguistic vitality potential while a score of (1) indicates low potential.
(4) Marriage between members of the same vernacular; children raised as monolingual
speakers of the vernacular
(3) Intermarriage practiced by some members of the speech community but children are
raised by parents or guardians within the vernacular community to speak vernacular
and possibly another language
(2) Intermarriage practiced; one or more parents maintain work outside vernacular
community and children raised for a period of time in household of grandparents
speaking vernacular
(1) Intermarriage practiced and family established outside vernacular community; speak an
LWC (language of wider communication or regional trade language) as their mother
tongue

Given the data gathered, we assigned a score of 2 on this scale for population and group dynamics.
1.3. Domains of language use
All respondents in our study were multilingual, speaking Khmu, Muang, and Standard Thai, though
we observed that the oldest respondents showed more limited standard Thai skills. Khmu was reported to be
the language used in the village between residents (even some non-Khmu spouses were reported to learn to
speak Khmu), and Muang for local commerce. Standard Thai is used in the domains of education, most
broadcast media, official functions, and is the only language of literacy.
Our research revealed that Standard Thai likely has increasing influence in the community and home,
domains in which Khmu has been the primary language used, due to two recent developments.

Firstly, Khmu pre-school children are being exposed to Standard Thai much earlier and more
intensively as a result of the opening of state-funded local daycare centers four years ago. Virtually all
children between the ages of two and five attend. The teachers in four of the five daycare centers are Khmu,
but are being encouraged by their sub-district administrator to speak Thai with the children.
Secondly, within the last five years television ownership has dramatically increased. According to
village leaders, five years ago almost no one owned a television, but now satellite television has become
available and our survey indicated that nearly 80 percent of respondents own a television and watch daily.
Below is Landweers (2000:9) ethnolinguistic vitality scale, which relates domains in which the
local language is used with potential for language maintenance:
(4) Home, cultural events, social events and other domains
(3) Home, cultural events, social events
(2) Home, cultural events where the vernacular is used, but is mixed with an outside lingua
franca or other local language(s)
(1) Home, where the vernacular is used but is mixed with an outside lingua franca or other
local language(s).

We assess the relative strength of Khmu language vitality in Chon daen to be a 3 on this scale
given that in other domains such as education, broadcast media, and commerce either Standard Thai or
Muang is the language used.
1.4. Access to a stable and acceptable income base within the locale
Landweer (2000) states that, One of the most common factors influencing a community to shift
from one language to another is that adequate work environments using their mother tongue do not exist for
their children (Landweer 2000:17).
The livelihood of the Khmu in Chon Daen is based primarily on subsistence hillside rice farming
and on gathering forest products. Some Khmu families in Chon Daen raise cows, pigs and or chickens on a
small scale as part of their livelihood base, and a few families have small fish farms. Nearly all of the
respondents to our survey questionnaire (89%) report they are involved in farming. The respondents who
reported student as their occupation (8%) help their parents on the family farm when they are free, bringing
total number of respondents involved in farming to nearly 100%.
Village leaders described two difficulties Khmu community members currently experience in trying
to provide for the needs of their families in the locale.
Firstly, the Khmu in Chon Daen do not have legal entitlement to the land they farm. The inability to
own land in the village locale causes stress and frustration, because without land rights Khmu have no
economic security, no means to acquire additional land needed to adequately support their families, and no
source of collateral to apply for loans. Within the last decade some of the land used by the Khmu was even
reclaimed as part of a national reforestation initiative. This particular land had been used for grazing cattle,
thus the loss of this land forced most residents to sell off their cattle. Cattle ownership had served as a sort of
emergency savings fund for many Khmu families.
A second challenge the Khmu face relates to the lack of a stable means of earning cash in the village
locale. Corn is being developed as a cash crop, but the market price is volatile. Many families accrued further
debt on their investment in corn crops in the year this study was conducted due to the low market price, the
costs of fertilizers and loss caused by rats and wild pigs. One leader lamented that all of the families in his
village had accrued large debt associated with the costs of agriculture and there was no way to earn cash
locally to repay these debts.
Landweer (2000:17) has developed a scale linking ethnolinguistic vitality to the availability of a
stable means of income generation in the local setting. This scale emphasizes the importance of having a

means of access to stable income bases in an environment that uses the local language. (A higher score of
the below scale indicates greater potential for language maintenance.)
(4) Stable and acceptable economic base where the vernacular is the code of choice
(3) Adequate dual economy where the language used is dictated by choice of economic
base
(2) Marginal subsistence economy requiring augmentation of the traditional means of
subsistence with non-vernacular, cash-based economic schemes
(1) Dependence on an economy requiring use of a non-vernacular

From the data we have gathered, our assessment is that the Khmu speech community in Chon Daen
rates a 1.5 on this scale. The issues discussed here have led to increasing debt and a growing dependency on
finding outside employment. Khmu families face an economic situation that depends on the use of Standard
Thai or a regional language such as Muang in order to secure wage labor in urban centers.
1.5. Social outlook regarding and within the speech community
According to Landweer (2000), the primary focus of inquiry regarding this vitality factor is, Is
there internal and/or external recognition of the language community as separate and unique within the
broader society? (Landweer 2000:15). Landweer explains that a strong ethnic identity can influence
language choice. In other words, the perception a group has of itself can be supportive or can undermine the
value associated with their language and ultimately their own use of the language (Landweer 2000:14).
Cultural elements that the Khmu in Chon Daen sub-district described as unique and important to
their identity included their language, traditional agricultural and healing ceremonies, rice-wine production,
and stories of the ancient cultural hero Ceuang. Khmu-style farming, trapping and gathering from the forest
were also mentioned as valued and uniquely Khmu.
Village residents are keenly aware that the Khmu are losing their distinctiveness, however. The
village leader of Ban Mai Chaidaen shared his insights as to why this is happening:
If you look at the various tribes [found in Thailandhe mentioned the Karen,
Lahu, , Hmong, Mien, Htin and Khmu], the Htin, Khmu and Khmu Lue seem to be most
flexible...discarding their customs and culture in order to follow the ways of others... Khmu
men no longer have traditional clothes...our women dont really wear the traditional clothes.

Several Khmu described the conflicting feelings and the stess that comes from living between
traditional and encroaching modern life-styles. The leader of Huay Klaep village articulated his outlook on
the times during an interview as follows:
We used to be a very small group that did not interact much with broader society
and its influences, but in the past 20 years the values related to materialism have crept in and
led to changes. In the past, girls did not really go out from the village to work. We used to
simply eat foods from the forest, but now people, especially teens, see the comforts and
luxuries of the outer society and want them; they do not want the old ways.
If society changes fast and we are not able to adjust ourselves rapidly enough, it will
be hard for us. Take car ownership, for example. If we do not have the resources to buy a
car and borrow money as is done in other parts of society, the payments are a burden and
hardship.
Modernizing brings benefits though. For example, in previous times we did not have
tin or tile for roofs, but we used leaves. Making a leaf-roof was a lot of work and after two to
three years it would leak and we would have to re-thatch. We have adopted this modern
development and benefitted. So, if we dont adopt these newer ways...it can bring suffering
as well (Village leader Sanit Saorungtoy).

We used two questions in our survey to probe community perceptions regarding attitudes of young
people toward Khmu language and culture. In response to the question, Are the young people proud of

being Khmu? 75% responded affirmatively, while several were unsure and only one interviewee responded
that young people were not proud of being Khmu.
A second question asked, Have you ever observed situations in which Khmu young people chose
not to speak Khmu to each other? Respondents related observations that when Khmu youth were in
Bangkok, outside the village, in a situation where outsiders are present or when at school they tended to
speak Thai or Muang. Several respondents felt that Khmu was not spoken because young people were
embarrassed to speak their mother tongue. Several respondents did note that in the village Khmu youth
speak Khmu together. One youth in his early twenties said that young people use a lot of slang that is really
a mixture of languages.
From these responses we conclude that Khmu youth still have a level of appreciation for their
language and culture, but are aware of and probably somewhat embarrassed by the lower social status of their
language when with speakers from a language and culture with higher prestige. Among Khmu peers in the
village, Khmu seems to be the language that is spoken.
As for the perspective of northern Thai outsiders in the surrounding area regarding the distinctives
of the Khmu, those we talked to knew little about the Khmu except that Khmu are not followers of Buddhism
but practice a spirit religion.
Ethnolinguistic vitality
factor
Significant changes or
observations reported
Overall future
language maintenance
impact (positive,
negative, or neutral)
Landweer
vitality scale
rating where
applicable
1. Intergenerational
language transmission
Children speaking K in the home
K the language of communication between
3 generations within households
positive Not applicable
Dramatic increase in young women
working outside village
Few K with K marriages
Attitude toward mixed marriages
increasingly neutral
negative
2. Population and group
dynamics
Few non-Khmu immigrants in the villages neutral
2
K the primary language of homes in village
(childrens first language)
K the primary language used in daily
village life
K the most fluent language for nearly all
respondents
positive
ST used exclusively in education; urban
workplace
Earlier exposure to ST at day care for 2-3
year olds
negative
3. Domains of language
use
M used for local commerce neutral
3
4. Access to a stable and
acceptable income
base within the locale
Lack of land rights
Market for cash crops unstable
Subsistence rice farming; large debt
Increased demand for consumer goods
through broadcast media exposure
negative 1.5
Positive attitude toward culture/language
and local language development
positive
Disappearing visible cultural distinctives;
flexibility
negative
5. Social outlook
regarding and within
the speech community
Youth desire and pursue modern lifestyle negative
2
Table 2: Summary of ethnolinguistic vitality factors

Abbreviations: Standard ThaiST; MuangM; KhmuK.

Landweers (2000:15) scale for rating a speech communitys social outlook in terms of its influence
on language maintenance or shift is presented below:
(4) Strong internal identity, high status or notoriety conferred by outsiders, with cultural
markers present
(3) Strong internal identity, neutral status or notoriety conferred by outsiders, with cultural
markers present
(2) Weak internal identity, neutral status or notoriety conferred by outsiders, with some
cultural markers present
(1) Weak internal identity, negative status or notoriety conferred by outsiders, with few if
any cultural markers present

Based on the data we gathered, we have assigned a score of 2 for this ethnolinguistic factor.
Summary of factors indicating potential for language maintenance or shift in Chon Daen Khmu
villages
Transmission of the language to children in the home is indeed the most positive indicator of
ethnolinguistic vitality. Overall, however, language and culture maintenance does seem to be threatened in
some significant ways. Table 2 below summarizes the ethnolinguistic vitality factors discussed and describes
the nature of the impact they may have on future language maintenance. Regarding the Landweer vitality
scale score, recall that 4 indicates high potential for language maintenance and 1 low potential.
2. Standard Thai literacy skills and practices
In our literacy survey we gathered data related to literacy skills in Standard Thai, and use of print
and broadcast media in order to inform potential local language development activities that may include the
production of literacy instructional materials, local language literature, and non-print media products.
2.1. Literacy skills
For the purpose of this study, we used the fairly broad definition of literate put forth by UNESCO:
A person is literate who can, with understanding, both read and write a short statement on his or her
everyday life (UNESCO 2006). We did not attempt to administer a literacy skills test, but used a self-
reporting evaluation instrument to gather data. Respondents to our literacy questionnaire were asked to
describe their current reading and writing abilities by choosing between the four skill levels described in
Tables 3 and 4 below. Note that reading and writing were treated as separate skills.
Reading skill level choice Definition of skill level
1. Cannot read I cannot read any words.
2. Poor I can read only a few words.
3. Fair I can read most words and sentences in books and newspapers; I read
somewhat slowly.
4. Very good I can read anything I want such as a newspaper, government forms,
agriculture or health materials without difficulty and fairly quickly.
Table 3: Skill level choices for reading skills

Writing skill level choice Definition of skill level
1. Cannot write I cannot write any words.
2. Poor I can write my name and a few other words.
3. Fair I can write simple sentences, but have trouble spelling many words.
4. Very good I can write a letter, story, report and fill in a form without difficulty.
Table 4: Skill level choices for writing skills

The respondents were grouped into two categories based on their self-assessments as follows:
1. Literate if reported their ability to read AND write as being Very Good or Fair
2. Not Literate if reported that they Cannot Read/Cannot Write AND/OR Read Poorly/Write
Poorly

Among younger cohort (ages 15-34), all of the respondents (18 of 18) rated their reading and writing
skills as either Fair or Very good and were therefore classified as Literate, though the women rated
their ability levels higher than the men overall.
Our findings were quite different among older cohort (ages 35 and older). Only four of 18 (or 22%),
were classified as Literate and only one of these was a woman.
2.2. Literacy practices
When we investigated the question,For what purposes do Khmu community members use literacy
skills? we learned that generally, most Khmu in Chon Daen do not depend on written materials to access
information, neither do many people read for entertainment or pleasure. There are several possible reasons
for this:
1) The majority of the respondents over 35 years old were farmers and the skills required to main tain
their livelihood were learned by doing and not dependent on literacy.
2) Access to newspapers and other types of reading material is quite limited. There were no reported
village libraries or reading rooms.
3) Other media forms are increasingly available and appealing. Nearly 80 percent of respondents watch
television daily and 70 percent report listening to the radio daily.
3. Attitudes toward language development
The prospect of developing a way to read and write Khmu was met with many enthusiastically
positive responses from village leaders, parents, community members and even Thai teachers and
administrators in the local schools. Several noted that other ethnic monirity groups around them had
developed writing systems and literature in their languages. However, some of the respondents were
skeptical as the whether it would be possible to develop a way to write the unique sounds of their Khmu
language.
The most commonly articulated perceived benefit of developing a writing system was that doing so
would help preserve the Khmu language and culture. Other perceived benefits were also mentioned and
included the following:
We could write and tell people our true feelings and others would know what we mean right
away.
People could write secrets to each other.
Khmu would be a complete language.
People who are not Khmu could learn Khmu.
We could write the words on paper for the Khmu spirit ceremonies.
If Khmu people could produce a written history of our people, written in our own language, we
would have a better understanding of who we are. We would have more confidence to try to
pull ourselves up [from a lower socio-economic status].
4. Perceptions of educators working in Khmu communities regarding the adequacy of Khmu childrens
Thai language skills for successful classroom learning
Because our study revealed that nearly all Khmu children in Chon Daen sub-district are monolingual
speakers of Khmu at the time they enter school, and there is a growing concern in Southeast Asia for the
welfare of students who must cope with learning through a language that is not their mother tongue (Kosonen

& Young 2009), we thought it important to learn about the educational experience of Khmu children, more
specifically asking the following questions:
Is there a perception among Thai educators working in Khmu communities that a language barrier
adversely affects the Khmu students ability to learn academic content?
Do local educators see value in incorporating a mother-tongue learning component into the local
curriculum?

We interviewed teachers and administrators at two schools, serving over 300 students, the large
majority of whom were Khmu. Key findings from our interviews include the following:
Teachers do experience significant difficulties communicating with the students in grades
Kindergarten through primary grade 2 (prathom 2).
Most teachers believe that a lack of Standard Thai speaking and listening skills adversely affects the
students educational outcomes, but several added that a language barrier is not the only factor that
may cause some children to struggle. One teacher who had taught young children in other minority
language communities suggested that a mother-tongue first curriculum would help the children learn
better during the early years of school.
Limited data on national exam scores (O-NET) and grade repetition rates indicate that Khmu
students are well behind many of their national peers regarding mastery of academic content (Miller
& Miller 2009).
The headmaster and several teachers expressed support for teaching Khmu culture and language as
one component of the school curriculum. The headmaster said he would implement a curriculum
immediately if he had one, but lacks the expertise and financial resources to develop such a local
language curriculum component.
Regarding informal language use by the students during breaks from the classroom, all the teachers
we interviewed reported that primary school children (through primary 6) speak Khmu with each
other.
5. Conclusions: Moving forward with language development
The purpose of this study was to inform a potential language development effort among the Khmu in
Thailand. In this section we will make specific recommendations regarding Khmu language development
based on these research findings. It should be noted that since this research was conducted, leaders from
these Khmu communities have taken the initiative to pursue orthography development that employs the
Standard Thai script. A recent partnership has been established between these communities, Mahidol
Universitys Institute for Cultures and Languages in Asia, SIL, and the Thailand Research Fund in order to
conduct community-based research leading to local literature development.
5.1. Now is the time to pursue Khmu language development
The Khmu are facing a rapidly narrowing window of opportunity in which to safeguard their
language. There is currently a vital, multi-generational community of speakers and yet the past decade has
brought some significant changes that are weakening the chain of language transmission to the youngest
generation. Fishman (1991), Crystal (2000), and Brenzinger et al (2003) all emphasize the foundational role
of orthography development, language documentation, local language literature production and facilitation of
teaching the language in school in safeguarding and strengthening threatened languages.
5.2. Develop a network of Khmu in Thailand
Because this study revealed that Khmu communities are fairly isolated from one another, it will be
important to develop a closer network so that the Khmu community at large in Thailand can contribute to and
benefit from resources that are developed. A community newsletter, website, and Khmu language radio
programming could help facilitate the building of closer ties between Khmu communities.

5.3. Facilitate Khmu literacy skills development
Our literacy survey indicated that there are both literate and non-literate adults in the community
who want to learn to read Khmu. Different literacy instructional approaches will need to be developed to
meet the needs of these two groups.
For those who can read Thai, self-study materials that employ a transfer skills approach (Waters
1998) would likely be sufficient to facilitate acquisition of Khmu language literacy skills.
For children or adults who do not have literacy skills in any language, literacy instruction will
require the participation of trained teachers and the development of materials intended to teach initial literacy
concepts and skills.
5.4. Develop relevant print and non-print Khmu language media
Research revealed that both the use and availability of reading material in these communities was
minimal. Thus a local language development program should include the development of literature that is
perceived to be interesting and relevant, as well as a means of making this literature readily accessible.
Further community research should be done to identify story-tellers, singers, musicians and those
with other special skills and knowledge so that these cultural forms and resources can be captured in both
print media and audio-visual formats such as video and photo journals.
Networking with local organizations (such as the local public health center) could facilitate
cooperative efforts to produce resources in the Khmu language that can have practical benefits in meeting
communication needs.
5.5. Support local language instruction for Khmu children in the formal education system
There is growing support in Thailand for using local languages in education. Most commonly, local
language and local culture appreciation are taught as a school subject, though there is also cautious support
for a local language-based multilingual educational approach in some contexts. For the Khmu children of
Chon Daen, both of these classroom applications would be appropriate.
The teachers in schools serving Khmu children gave clear testimony that in the early grades the
Khmu children have difficulty understanding Standard Thai, the language of classroom instruction. An
educational program that allows the students in the early grades (kindergarten and early primary) to begin
learning in their mother tongue and gradually bridge into Standard Thai-based instruction would likely
increase students comprehension of academic material and improve educational outcomes (Kosonen and
Young 2009). Continuing to incorporate the teaching of Khmu as a subject, then, for older students would be
important step toward safeguarding the Khmu language by fostering appreciation for their local language and
promoting its use in new domains (Crystal 2000).
5.6. Develop new avenues for local income generation
Consult with outside experts in agriculture and microenterprise to pursue developing means of local
income generation that are more viable and stable than current cash crop ventures. The movement of young
people out of Khmu communities and into urban areas will likely increase if new avenues of local income
generation are not developed.
6. Recommendations for further research
Sociolinguistic research and documentation needs to be extended to include other Khmu
communities in Thailand to determine whether their situation is similar to that of the Khmu in the
communities that participated in this study.

REFERENCES
Aguettant, Joseph L. 1996. Impact of population registration on hilltribe development in Thailand. Asia-
Pacific Population Journal 11:4. 47-72.http://www.uniscap.org/esid/psis/population/journal
/1996/v11n4a3.htm (Accessed 17 June 2004).
Benson, Carol and Kimmo Kosonen. 2010. Language-in-education policy and practice in Southeast Asia in
light of the findings from Ethiopia. In K. Heugh & T. Skutnabb-Kangas (eds.). Multilingual
Education Works. New Delhi: Orient Black Swan (Forthcoming).
Brenzinger, Matthias et al. 2003. Language vitality and endangerment: By way of introduction. Paris:
UNESCO Ad Hoc Expert Group on Endangered Languages. http://www.linguae-
celticae.org/dateien/ UNESCO_Language_vitality_and_endangerment.pdf. (Accessed 19 October
2009).
Chazee, Laurent. 1999. The peoples of Laos. Bangkok: White Lotus Press.
Cholthisa Ong-arj. 1988. A phonology of Khmu Lue at Nan province. Salaya: Institute of Language and
Culture for Rural Development, Mahidol University M.A. thesis.
Cooper, Michael and Susan Cooper. 1998. A preliminary phonological analysis of Huay Jo Khmu
(Technical Paper No. 38). Chiang Mai: Payap Research and Development Institute and the Summer
Institute of Linguistics.
Cooper, Michael and Susan Cooper. 1999. Tri-lingual phrasebook: Khmu -Thai -English (Technical Paper
No. 41). Chiang Mai: Payap Research and Development Institute and the Summer Institute of
Linguistics.
Crystal, David. 2000. Language death. Cambridge: The Cambridge University Press.
Fishman, J.A. 1991. Reversing language shift. Clevedon: Multilingual Matters.
Gordon, Raymond G., Jr (ed.). 2005. Ethnologue: Languages of the world. 15th Edition. Dallas, TX: SIL
International. http://www.ethnologue. com/show_country.asp?name=TH (Accessed 30 September
2009).
ILCRD Research. 2008. Endangered languages and cultures. Institute of Language and Culture for Rural
Development. Salaya, Thailand: Mahidol University. http://www.lc.mahidol.ac.th/en/research-
groups-endangered.php (Accessed 15 June 2009).
Kosonen, Kimmo and Catherine Young (eds). 2009. Mother tongue as a bridge language of instruction:
Policies and experiences in Southeast Asia. Bangkok: SEAMEO.
Landweer, M. Lynn. 2000. Indicators of ethnolinguistic vitality. Dallas: Notes on Sociolinguistics 5:1.5-22.
Lewis, M. Paul. 1996. Language maintenance in seven Kiche communities. SIL Electronic Working Papers
1996-002, July 1996. http://www.sil.org.silewp/1996/002/silewp_1996-002.html
Lindell, Kristina. 1974. A vocabulary of the Yuan dialect of-the Kammu language. Acta Orientalia 36.191-
207.
Lindell, Kristina, Jan-Olof Svantesson and Damrong Tayanin. 1981. Phonology of Kammu dialects. Cahiers
de Linguistique Asie Orientale 9.45-71.
Miller, Michelle and Timothy Miller. 2009. A sociolinguistic and literacy study of Khmu communities in
Thailand: toward the process and products of language development. Chiang Mai: Payap University
Linguistics Institute.
National Statistics Center of the Lao PDR. 2007. http://www.nsc.gov.la (Accessed 19 October 2009).
Preisig, Elisabeth. 1990. The Kmhmu' orthography: Dialect of Xieng Khouang, Luang Prabang and Sam
Neua. Unpublished manuscript.
Smalley, William. 1961. Outline of Khmu structure. An essay of the American Oriental Society No 2, ed. by
Edward H. Schafer. New Haven: American Oriental Society.
_____. 1994. Linguistic diversity and national unity: Language ecology in Thailand. Chicago: University of
Chicago Press.
Suksavang Simana, Somseng Sayavong and Elisabeth Preisig. 1994. Kmhmu-Lao-French-English
dictionary. Vientiane, Lao P.D.R.: Ministry of Information and Culture, Institute of Research on
Culture.
Suksavang Simana and Elizabeth Preisig (English). 1998. Kmhmu livelihood: farming the forest. Vientiane,
Lao PDR: Institute for Cultural Research, Ministry of Information and Culture.
Supatra Jirananthanaporn. 1988. An experiment using Thai orthography with Khmu phonology at Ban Nam
Pan in Nan. Thailand: Mahidol University M.A. thesis.
Suwilai Premsrirat. 1987. Khmu, a minority language of Thailand. Pacific Linguistics: Series A, No.75.
Department of Linguistics, Research School of Pacific Studies, The Australian National University.

_____. 1990. Khmu traditional medicine and Khmu-Thai medical conversation. Salaya, Thailand: Institute
of Language and Culture for Rural Development, Mahidol University.
_____. 1993. Thai-Khmu-English dictionary. Salaya, Thailand: Mahidol University, Institute of Language
and Culture for Rural Development.
_____. 1998. Phonological variation and change in the Khmu dialects of northern Thailand. Mon-Khmer
Studies (MKS) 29.57-69.
_____. 2001. Tonogenesis in Khmu dialects of Southeast Asia. Mon-Khmer Studies (MKS) 31.47-56.
_____. 2002. Thesaurus of Khmu dialects in Southeast Asia. Salaya, Thailand: Mahidol University, Institute
of Language and Culture for Rural Development.
Svantesson, Jan-Olof, Damrong Tayanin and Kristina Lindell. 1994. Khmu-Lao dictionary. Vientaine, Lao
PDR: Ministry of Information and Culture.
Svantesson, Jan-Olof and David House. 2006. Tone production, tone perception and Kammu tonogenesis.
Phonology 23.309333.
Tehan, Thomas M. and Ramzi W. Nahas. 2009. Mpi present and future: Reversing language shift. Mon-
Khmer Studies (MKS) 38.87-104.
UNESCO 2006. Education for All Global Monitoring Report 2006. http//www.unesco.org/education/
GMR2006/full/chapt6_eng.pdf. (Accessed 1 March 2009).
Vail, Peter. 2006. Can a language of a million speakers be endangered? Language shift and apathy among
Northern Khmer speakers in Thailand. International Journal of the Sociology of Language 178.135-
147.

Osborne, Anne R. 2011. Multiple Uses of the Pronoun de in Kmhmu'. In Sophana Srichampa and Paul Sidwell (eds.)
112
Multiple uses of the pronoun de in Kmhmu'
R. Anne Osborne
SIL International
Kmhmu' personal pronouns have a singular, dual and plural distinction, with gender differentiation
in the second and third person singular forms. As well as these forms there is also an unspecified pronoun de.
Unlike the other personal pronouns, de has no semantic constraints with respect to person, gender or number.
It exhibits a range of meanings depending on its context. From the speaker's point of view, what motivates
the choice of de over one of the specific pronouns? How do the hearers know who de is referring to? Are
the contextual factors involved in interpreting the meaning of de purely pragmatic, or can syntax help? This
paper explores the syntactic and pragmatic constraints that govern the use and interpretation of de.
The data for this paper comes from the Kmhmu' spoken in Vientiane province of the Lao PDR.
Many people have helped with collecting, translating and analysing texts. In particular I would like to thank
Ajarn Suksavang Simana', Elisabeth Preisig, Ajarn Sosavanh Silaphet, Mrs Tan Ounpachanh and Miss Pang
Vilay.
Dr Suwilai (1987:33) and Suksavang et al. (1994:166) describe de as a reflexive pronoun. Van den
Berg (1988:5) says it does not have a reflexive meaning, but occurs in certain dependent clauses where it is
co-referential with the subject of the main clause. Suksavang et al. (1994:166) also report a co-referential
meaning and observe that de can replace all other pronouns, and can also mean alone. Building on these
findings, this paper presents four structurally defined uses of de, one of which has multiple pragmatic
functions.
Syntactically Defined Uses of de
These four uses of de occur in distinctive syntactic contexts. Where there is an antecedent reference
in a clause, de has a co-referential function. Where there is no antecedent, de is interpreted as a first person
meaning that is unspecified for number. As the final element in a clause, de means alone'. When it occurs in
the verb complex, de combines with a verbal element to add aspectual or adverbial information to the main
verb.
Co-referential function
In a well-formed sentence de is used anaphorically for any pronominal co-reference within a clause
and across some clause boundaries. This syntactic context requires an antecedent reference for de. In (1), de
is co-referent with the antecedent first person singular pronoun o.
(1)
l o
i
jt j mkn de
i
/
#
o
i

and 1sg stay with aunt CO-REF/ 1sg
And I stayed with my aunt
93
.

93
In the Kmhmu' possessive, possessors follow the possessed entity.
R. Anne Osborne 113

Although use of the specific pronoun is not ungrammatical in this context, it is not considered to be
good Kmhmu.
Co-reference often occurs in a possessive and may be used in inalienable possession such as kin
terms (see (1)), and body parts, as shown in (2) in the possessive hr m de his heart/mind'.
(2)
i
t hr m de
i
l l
3sgm so_then set_up heart CO-REF good good
So he made up his mind firmly.

In alienable possession also, de may refer to the possessor, as shown in (3), where the possessor in
om de her water' is co-referent with the subject n 3sgf .
(3)
n
i
k om de
i

3sgf so_then carry_on_shoulder water CO-REF
Then she carried her water on (her) shoulder

When de has a co-referent function, the antecedent is usually the subject of the clause, as shown in
(4) where de is co-referent with j nj that woman'.
(4)
j nj
i
n knn m
j
de m uk de
i/ *j

woman that give young_person get hat CO-REF
That woman gave the young person her hat.

A non-subject antecedent is also possible, though unusual. An example is shown in (5) where de is
co-referent with knn m young person, the non-subject argument in the clause.
(5)
j nj
i
n knn m
j
hk de m uk de
j/ *i

woman that give young person nevertheless get hat CO-REF
That woman gave the young person his/her hat.

In order for de to be understood as referring to the non-subject argument, the conjunction hak
nevertheless must be inserted in the clause. Examples (4) and (5) demonstrate the effect of hak
nevertheless in changing the co-reference relation.
The domain of co-reference for de is a clause. The clause boundary includes embedded clauses
within a matrix clause, subordinate adverbial clauses and tail-head linkage clauses in a narrative. It does not
extend across co-ordinate clause boundaries.
An example of an embedded clause is seen in (6), where de is the subject of the complement clause
de ci han kni I will die now, and is co-referent with the subject of the matrix clause o 1sg.
(6)
o
i
de
i
ci han kni
1sg fear CO-REF IRR die now
I am afraid (that) I will die now.

Co-reference with the object of the matrix clause requires use of a specific pronoun, as shown in (7)
where 3sgm' is co-referent with kn child the object in the matrix clause.
(7)
o
i
mk kn
j
n
j/ *i
jh
1sg tell child IMP 3sgm go
I told the child to go.
114 Multiple Uses of the Pronoun de in Kmhmu

If de is used in this context, as shown in (8), the sentence is ungrammatical because it has no
meaningful interpretation.
(8)
* o
i
mk kn
j
n de
*i/ *j
jh
1sg tell child IMP CO-REF go
I told the child ?

An example of co-reference in an adverbial clause is seen in (9), where de: is the subject of the
clause p de i h i rm kn ln m so that he would become richer than before and co-referent
with the subject in the main clause tj his older brother.
(9)
tj de sun kp tut trul p
older_sibling 3sgm get garden with fig_tree PURP

de i h i rm kn ln m
CO-REF IRR have IRR rich AUG more_than old
His older brother got the garden and the fig tree, so that he would become richer than before.

An example of co-reference in a tail-head linkage is seen in (10). In the second sentence, when the
clause is repeated, typically the subject is a null reference, and de is co-referent with it.
94

(10)
l o
i
jt j makn de
i

and 1sg stay with aunt CO-REF
And I stayed with my aunt.

i
jt j mkn de
i
mkn o
i
m rk
1sg stay with aunt CO-REF aunt 1sg NEG love
(I) stayed with my aunt, (but) my aunt did not love (me).

Co-reference does not extend across co-ordinate clause boundaries, but rather an NP or specific
pronoun is required to identify the same referent. This is seen in (11), where de is used as a co-reference in
the first clause, but in the 2
nd
clause the specific first person singular pronoun o must be used to identify the
same referent.
(11)
o
i
jt j mkn de
i

1sg stay with aunt CO-REF

nj mkn o
i
/*de m rk
but aunt 1sg/CO-REF NEG love
I stayed with my aunt, but my aunt didn't love (me).

In Kmhmu, clauses may be juxtaposed without any overt marking of the hierarchical relationship
between them. The scope of the co-referential function of de has implications for identifying clause
boundaries and the nature of the relationship between clauses. Where de continues to be co-referential across
clause boundaries, it is indicative of a subordinate hierarchical relationship between clauses. Where it does
not have scope over adjacent clauses, this marks a co-ordinate relationship.

94
If this construction is analysed without a null subject, then a tail-head linkage is a case where the antecedent
of de is outside the clause.
R. Anne Osborne 115

Unspecified first person meaning
In the second of the four structurally defined uses of de, where there is no syntactic antecedent, de
points to a referent in the speech situation or the cultural context. It evokes a set of referents that includes the
speaker and is not specified as to number.
An example of this unspecified first person meaning is seen in (12) where de occurs in the second
sentence. There is no antecedent to de in this clause or the preceding one and de is interpreted as having the
non-specific meaning one. This would include the speaker and anyone who might climb the mountain.
(12)
o mok ni
so_then far oh! mountain this
Oh! (it) is far away, this mountain.

m de kndrum mok d n
when unspecified climb underneath mountain at up_there
pmn sm si l k ni l
about three four Clf_kms this PRT
When one climbs the lower slopes of the mountain up there, (it is) about three (or) four kilometres.

In another context where there is no antecedent, de combines with the noun on person in the
construction on de we people. An example is shown in (13).
(13)
n rwj ni u pok me pok
COND tiger this want bite INDEF 3sgn so_then bite
mh on de t m h j m bn
eat person unspecified do INDEF with 3sgn NEG can
If the tiger wanted to attack anyone, to eat anyone, then it attacked (and) ate (them); we people could not do
anything to it.

In this context the noun on person, people would identify a global generic referent, i.e. the whole
human race. Combined with de it refers to people, including the speaker, within the wider relevant
community i.e., those who were in the area affected by the tiger.
95

It is not clear whether on de is a compound or a noun phrase. The only element which may be
inserted between on and de is the possession marker de, which then changes the meaning from we
people to our people. Other elements, such as an adjective, relative clause, demonstrative, or indefinite
pronoun may not be inserted between on and de, as shown in (14).
(14)
on de de t m h j m bn
person POSS unspecified do INDEF with 3sgn NEG can
our people could not do anything to it.

*on n m de t m h j m bn
person big unspecified do INDEF with 3sgn NEG can
we big people could not do anything to it.

95
on de is less restrictive than on i we (specific group) people, which includes the speaker and their
family or a specific group identified in the discourse context.

*on jt ket ni de t m h j
person live district this unspecified do INDEF with
m bn
3sgn NEG can
we people (who) live in this district could not do anything to it.

*on ni de t m h j m bn
person this unspecified do INDEF with 3sgn NEG can
we this people could not do anything to it.

Thus it seems that on de may be a compound. Kmhmu' speakers think of the construction as two
words and it is difficult to come to a definitive answer on this.
In other speech situations, de identifies a more specific first person referent, either singular or plural.
An example of a plural referent is shown in (15). There is no antecedent reference for de in this sentence. It
is interpreted using cultural knowledge as a first person plural meaning, the family of the speaker, i.e. ku de
our village.
(15)
r ru t d wi rt d ku
road rough starting at Vientiane arriving at village
de ni
unspecified that_one
The road was rough starting at Vientiane (and) arriving at our village.

An example of a singular referent is shown in (16). Again there is no antecedent reference for de in
this sentence. It is interpreted using the context of the discourse as a first person singular meaning.
(16)
ho jo o lw sh i m l b
and_then father 1sg say COMP IRR NEG good eat
oms jr sh de srm
tea because unspecified have_fever
And then my father said, (It) would not be good to drink, tea, because I had a fever.
Alone
A third syntactic context provides another interpretation of de. When de occurs as the final element
in a clause, it can mean alone' or by oneself
96
. This is seen in (17).
(17)
bt i jh de
turn this_one 3sgm so_then go oneself
Then (at) this time he went alone.

This meaning is made clearer to the audience by the optional insertion of the counter expectation
marker hk nevertheless, as shown in (18).

96
There is no sense of an emphatic meaning here, where a referent is re-iterated to emphasise that it was this
person and not another who went, such as in English He himself went. Nor is there a reflexive meaning, which
typically involves a single referent being the actor and undergoer of the same action, e.g. He saw himself (in the
mirror). It is closer to the English use of the reflexive pronoun in a preposition phrase e.g. by himself.
R. Anne Osborne 117

(18)
bt i hk jh de
turn this_one 3sgm so_then nevertheless go oneself
Then (at) this time he went alone.

Where the subject is plural, this sense of de can also be used, see (19).
(19)
plj n n (hk) jh de
release CAUS 3pl nevertheless go oneself
(We) let them go alone.

This usage is seen less frequently than the co-referential or unspecified first person meanings.
Particle in the verb complex
The fourth syntactic context in which de is found is in the verb complex. In this context, de appears
to have an argument re-instantiation function rather than a referential function. Some speakers do understand
it as co-referential with the subject of the clause, but others see it as having no meaning of its own, but
merely necessary for a well-formed sentence. Therefore this use is included in this paper not so much
because de here is acting as a pronoun, but for the sake of completeness of the characterisation of
grammatical contexts in which it may be found.
In a Kmhmu' clause the verb complex is identified by its position in relation to the conjunction
so_then, which, when it occurs, always follows the subject and precedes the verb complex. When de
appears in this context with , it always follows and so is part of the verb complex. Within the verb
complex, de combines with a verbal element and precedes the main verb, adding components of meaning to
it. Five different verbal elements are seen with de, giving aspectual and adverbial information about the
action or state of the main verb.
bn de PAST
When used as a main verb bn means to achieve, to accomplish, to reach. In combination with de
in the verb complex, it signifies past action of the following main verb; something which has been achieved
or accomplished. In (20), bn de marks the construction of the rice field as an event accomplished in the
past.
(20)
moj d hh i bn de h r e
one Clf_times PRT 1pl PST construct rice_field
One time ok, we made an upland rice field. (or we got to make)

When used in this context de can be omitted without any loss of meaning, as shown in (21).
(21)
i bn h r e
1pl PST construct rice_field
We made an upland rice field. (or we got to make)

Negation of bn de indicates that an action was not achieved, as shown in (22), where m bn de
k om means didn't get to drink water.
(22)
t nw m m bn de k om
do whatever so_then NEG PST drink water
Whatever (he) did (I) didn't get to drink water.

t de RESULT
When used as a main verb t means to set up, to arrange, to commence. In combination with de
in the verb complex, it signifies an action resulting from the event in the previous clause. In an example from
a reported narrative about a bear hunt shown in (23), one of the hunters is attacked by the bear. His friend
goes to his rescue and takes out his knife and advances on the bear. As a result of this action the bear releases
the man. This is marked by t de in the result clause t de plj then it released (him).
(23)
j t mit d ua d i
3sgm but_then draw_out knife at waist at this_one
But then he drew out a knife at (his) waist here.

jh r t de plj
go approach 3sgn then release
(As he) went (and) approached, then (as a result) it released (him).

The conjunction can be inserted after the subject and before t de as shown in (24).
(24)
t de plj
3sgn so_then then release
So then (as a result) it released (him).

It is possible to omit de: with no obvious change in meaning, as shown in (25).
(25)
t plj
3sgn then release
Then it released (him).

When de is omitted, however, the inclusion of produces an ungrammatical sentence, as shown
in (26).
(26)
* t plj
3sgn so_then then release
So then it released (him).

Negation of t de is not possible, probably for semantic reasons.
p de almost, just about
When used as a main verb p means to be enough. In combination with de in the verb complex,
it adds an aspectual meaning to the main verb and signifies a state that is about to be achieved or has almost
come about. In (27) p de is used to add aspectual meaning to the adjectival state verb bh to be light.
(27)
i rh jh d r e n bri p de bh
1pl get_up go at rice_field when environment almost light
We got up (and) went to the rice field when it was almost light.

Omission of de from this construction is ungrammatical, as shown in (28).
(28)
* bri p bh
environment almost light
It was almost light.
R. Anne Osborne 119

Insertion of the conjunction is grammatical, as shown in (29).
(29)
bri p de bh
environment so_then almost light
It was almost light.

Negation of p de is not possible, perhaps due to semantic constraints.
m n de as if
When used as a main verb m n means to be like, to be the same as. In combination with de in the
verb complex, it signifies an event or state which appears to be true, as shown in (30).
(30)
ur i m n de i lm
stew this_one as_if IRR delicious
This stew looks as if (it) will be delicious.

Omission of de produces a grammatical sentence with no change of meaning, as seen in (31).
(31)
ur i m n i lm

The indefinite pronoun m someone, something can be substituted for de with no change in
meaning as shown in (32).
(32)
ur i m n me i lm

The fact that the indefinite pronoun can substitute for de gives some support to the idea that de
retains some of its referential functions in this context.
r j de together, with one another
When used as a main verb r means to lead, to accompany and j means together. In
combination with de in the verb complex, r j de generally precedes an action verb and signifies that a
group of people performed an action together, adding the meaning of accompaniment to the main verb. In (33)
r j de precedes the main verb hj kr to laugh at and indicates that the other people all joined
together to laugh at them.
(33)
l bri r j de hj kr sn
and other_people so_then together laugh_at 3du
And other people together laughed at them.

Omission of de: is grammatical and brings no change of meaning, as seen in (34).
(34)
l bri r j hj kr sn
and other_people so_then together laugh_at 3du
And other people together laughed at them.

Pragmatically Defined Uses of de
In the syntactic context where there is no antecedent reference and de is interpreted as having an
unspecified first person meaning, it may also communicate other meanings depending on the pragmatic
context. These meanings include marking events off the storyline of a narrative, mitigating the force of a
speech act and communicating friendship or intimacy.
Marking events off the storyline
In a personal narrative, de can be used with the first person meaning to mark events off the storyline
such as the author's comments on these events, or actions that are out of chronological sequence such as a
flashback.
An example of marking author comments is shown in (35). The author tells of an experience where
she is offered an ice-cream for the first time in her life. Her father asks her would she like one and she says
yes. In relating her response, an event on the storyline, she uses the specific pronoun o 1sg; ho o lw
b Then I said, (I) do,. In the subsequent 4 clauses she goes on to comment on her experience. In these
clauses she uses the unspecified pronoun de repeatedly; de m j b I had never eaten (it), de lw b
I said (I) did (want to eat it), de dm I believed. When she resumes the events of the story, she reverts to
the specific pronoun o 1sg; o lt lt jh I licked (the ice-cream), licked (and) went along.
(35)
ho o lw b nj de m j b moj
and_then 1sg say eat but unspecified NEG ever eat not_one
bt m j b moj d
Clf_turns NEG ever eat not_one Clf_times
Then I said, (I) do, but I had never eaten (it) once; (I) had never eaten (it) even once.

de lw b de dm sh i lm
unspecified say eat unspecified believe COMP 3sgn IRR delicious
I said (I) did (want to eat it), I believed that it would be delicious.

o lt lt jh ho jo o jh kl
1sg so_then lick lick DIR and_then father 1sg walk DIR before
I licked (the ice-cream), licked (and) went along, and my father walked along in front.

In (36) there is an example of marking events that are off the storyline because they are out of
chronological sequence i.e. a flashback. In the first sentence the storyteller uses the specific first person
singular pronoun o to refer to herself in an event on the storyline, njm l jo o The doctor then
scolded my father. As she relates events from a flashback in the next two sentences, she uses de to refer to
herself; de srm I had a fever, de n jh l I still went out to play, de srm I had a fever,
de mh kn I was a small child. When she returns to the storyline events, she resumes using the
specific pronoun o; seh o de om tle (They) put saline into me.
(36)
njm l jo o de sh jo o i
doctor so_then scold father 1sg get COMP father 1sg stupid
The doctor then scolded my father that my father was stupid.

jr sh m n jt d ku de srm si
because when yet stay at village unspecified have_fever today
pmi m srm srm bp trw m
tomorrow NEG have_fever have_fever type interval day
Because when (we) were still staying in the village I had a fever one day, the next day (I) didn't have a fever;
(I) had a fever alternate days.

R. Anne Osborne 121

si de n jh l pmi de
today unspecified so_then yet go play tomorrow unspecified
srm jr de mh kn srm ra lw
have_fever because unspecified be child small fever cease already
j jh l nw n
but_then go play type that
Today I still went out to play, tomorrow I had a fever, because I was a small child, (when) the fever had
ceased (I) went out to play, like that.

mh dni n lw sh mh srm pt
be like_that 3pl so_then say COMP be malaria
So (because it) was like that, they said (it) was malaria.

seh o de om tle on bn hok dj
put_in 1sg get saline until achieve six Clf_bags
(They) put saline into me, until (it) reached six bagfuls.

The default reference pattern for events on the storyline with the same subject as the previous clause
is a zero reference (Osborne 2009:95). The unusual referencing pattern seen in (35) and (36) with a switch
from the specific pronoun to repeated use of the unspecified pronoun de is a device for marking clauses
which are off the mainline of events.
Mitigating effect
In this same syntactic context where there is no antecedent reference, but in a different pragmatic
context, the unspecified first person meaning of de may be used to signal mitigation of the emotive force of a
speech act.
An example of this is seen in (37), which occurs later in the same story as (35). Having tried the ice-
cream and found it unpleasant, the author throws it away. When her father discovers this he scolds her and
addresses her using the pronoun de to soften the force of his speech.
(37)
b n de b de l pi
3pl give generic eat generic PRT discard 3sgn
Someone gives one (some) to eat, (and) one throws it away.

In all previous conversations with her the father has used the specific pronoun b 2sgf to address
her. The change to de signals a mitigating of the accusatory force of the second person pronoun. Because the
use of de includes the speaker, it necessarily lessens the accusatory impact of the rebuke. A similar effect in
English might be achieved by using first person plural instead of second person singular, e.g. When
someone gives us an ice-cream, we dont throw it away.
Signalling friendship and intimacy
Another conversational use of de, where there is no antecedent reference, is as a vocative to signal
friendship.
The pragmatic context for this use is specific to female speakers who are close friends and age mates.
It does not substitute for a specific pronoun, but rather for the name of the other person and may be
interpreted as meaning friend, as shown in (38).
(38)
de jh l tlt jh
friend 1du go play market go
Friend, (let) us go to the market, ok?

In another pragmatic context, that of close male-female relationships, de is used as a form of
intimate address. Either partner will address the other as de and refer to themselves as bri, which in most
contexts means other people. An example of this is shown in (39), where a young woman speaks to her
fianc on the phone of how she misses him. She uses de to address him and refers to herself as bri.
(39)
de sre bri oh
unspecified so_then think_about other_people PRT
bri sre de
other_people think_about unspecified
Are you thinking of me? I am thinking of you (all the time).
Relationship Between the Uses of de
There is a semantic core that is common to at least three of the four syntactically distinct uses of de
described in this paper. That is the element of co-referentiality. In the Co-referential function, the
Unspecified first person meaning and Alone, de identifies a referent that has already been supplied either
syntactically or from the speech situation or cultural context. Thus these forms are clearly semantically
related and are probably one word with different uses rather than homophonous forms.
The exact function of the fourth use, Particle in the verb complex, is not yet clearly defined. Unlike
the other three uses which are associated with reference, it is associated with the world of events. Thus it is
not yet clear whether this is another use of the same word or merely a homophonous form.
Of the three related uses, there is no synchronic evidence to point to which use was historically prior
to the others. In uses 1 and 3, Co-referential function and Alone, the co-referent is syntactically available,
while in use 2, Unspecified first person meaning, it is not. One could characterise the Unspecified first
person meaning as the more general interpretation which holds whenever the conditions for the other two
syntactically constrained uses are not met. Also, this use has a number of wider pragmatic interpretations and
perhaps on these grounds it might be hypothesised to be prior to the other uses. Shorto (2006) reconstructs a
proto-Mon-Khmer *e reflexive pronoun which has reflexes in Palaungic, Bahnaric and Nicobarese in
addition to Kmhmu. Further research in the documented history of related languages may provide evidence
for prior occurrence of the syntactic co-referent usage or the unspecified usage. What is clear from this study
is that speakers have specific ways to prevent ambiguity in meaning between the various uses of de. It is
these syntactic and pragmatic constraints that are the focus of this paper.
Conclusion
Four syntactically defined uses of the unspecified pronoun de have been characterised. Where there
is an antecedent reference, de has a co-referential function with scope within a main clause, over an
embedded clause, an adverbial clause or a tail-head linkage clause, but not over a co-ordinate clause. The
scope of the co-referential function of de has implications for identifying clause boundaries and the nature of
the relationship between clauses.
Where there is no antecedent reference, de has an unspecified first person meaning which is
identified from the speech situation or cultural context. In certain pragmatic contexts, this unspecified first
person meaning of de may be interpreted as marking events off the storyline of a narrative discourse,
mitigating the emotive force of a speech act, or signalling friendship or intimacy.
When de is the final element in a clause, it means alone. In the fourth syntactic environment, as
part of the verb complex, de occurs with other verbal elements to add aspectual or adverbial components of
meaning to the main verb.
Although this function of de in the verb complex and its relation to the other uses of de is not yet
clearly defined, the first three uses of de share the common semantic core of co-referentiality and thus are
R. Anne Osborne 123

taken to be different uses of a single word. Their use is governed by syntactic and pragmatic factors that
allow speakers to unambiguously identify referents and also communicate speaker attitude.
References
Berg, Ren van den, with the assistance of Mark Taylor. 1988. Some Khmu' particles A preliminary analysis.
Anduze, ms.
Osborne, R. A. 2009. Who's who in Kmhmu': Referring expressions and participant identification in selected
Kmhmu' narrative texts. Chiang Mai: PYU thesis. http://www.sil.org/silepubs/Pubs /52451/ Osborne
_WhosWhoKmhmu_52451.pdf
Shorto, Harry L. 2006. A Mon-Khmer Comparative Dictionary. Canberra, Pacific Linguistics 579.
Suksavang Simana, Somseng Sayavong & Elisabeth Preisig. 1994. Kmhmu' Lao French English
dictionary. Vientiane: Ministry of Information and Culture, Institute of Research on Culture.
Suwilai Premsrirat. 1987. A Khmu grammar. Pacific Linguistics Series A No. 75, Papers in South-east
Asian Linguistics 10. 1-143.

Paillard, Denis. 2011. About aj in Contemporary Khmer. In Srichampa and Sidwell (eds.) Austroasiatic Studies:
papers from ICAAL4. Mon-Khmer Studies Journal Special Issue No. 2. Dallas, SIL International; Salaya, Mahidol
University; Canberra, Pacific Linguistics. pp.124-137.
124
About aj in Contemporary Khmer.
Denis Paillard
Laboratoire de linguistique formelle
Universit Paris Diderot
GIVE-verbs
97
have given rise to numerous studies in various languages (cf. Newman: 1996, 1997
for a bibliography). In most languages, GIVE verbs have other uses than that described by Newman as the
act of giving. Heine & Kuteva (2002) mention the following uses: benefactive, dative, causative,
purpose. In some languages, GIVE-verbs are used as a passive form (cf. in particular Paris: 1998, Yap &
Iwasaki: 1998, 2003). It should be noted that depending on the languages, GIVE verbs have widely differing
uses, which suggests that the semantic core of the word cannot be reduced to that of transfer. Just compare
for instance the uses of give in English and donner in French
98
.
Our study of the uses and values of aj in Khmer will definitely not consider the transfer value as
the basic one. Grounding on an inventory as complete as possible of all these values in contemporary Khmer,
we bring forward a characterization of the semantic identity of aj which makes it possible to account for
the range of all its uses, showing what they have in common and in what they differ. Such a characterization
will neither resort to the notion of grammaticalization nor that of desemantisation. On this point, our analysis
differs from other studies dealing with GIVE-verbs in South-East Asian languages, such as Khmer (Bisang,
1996) Tha (Rangkupan, 2005, Thepkanjana & Uehara, 2008) or Mon (Jenny (2006).
1. Inventory of the uses of aj
Our starting point is not a semantic one (we do not base on the transfer value) but a syntactic one,
namely the various syntactic constructions where aj is to be found. We present ten groups of uses of aj ;
each type of use is illustrated through a series of representative data and corresponds to specific syntactic
properties. The terms used to refer to the various uses are purely conventional and are presented with the
corresponding syntactic patterns.
1.1. Transfer: N
1
aj N
2
N
3
/ N
1
aj N
2
V N
3

1a) koat aj baj cka ma pnman
3sg aj rice dog hour how many
At what time does he give the dogs their meal?
1b) koat aj baj cka si ma pnman
3sg aj rice dog eat hour how many
At what time he will give the dogs their meal? (they are starving)

97
We use here GIVE as a generic term referring to verbs which in various languages cans express the notion of
transfer. This does not in any case mean that those verbs are synonyms of give in English.
98
From a diachronic point of view, cf. the text by E. Benveniste about the notion of gift in Vocabulaire des
institutions indo- europennes (1993). Benveniste shows that the bases giving rise to give and donner correspond to quite
different representations. About yaru in Japanese, let's mention Oguma (2006).
Denis Paillard 125

2a) campuh kai smn koat aj kam
ka
as regards work construction 3sg aj worker
muj taj pram rj ril
one day five hundred riel
As regards the construction works, he gives the workers 5000 riels a day (workers salary)
2b) campuh kai smn koat aj t kam
ka

as regards work construction 3sg aj go worker

muj taj pram rj ril

one day five hundred riel

As regards the construction works, he spends 5000 riels a day for the workers
(managers expenses for the salaries)
As shown by examples (1) and (2), it is possible to have aj either alone or combined with another
verb (namely t go in most cases). With aj alone, the Ns stand for arguments ((agent, object,
beneficiary) of aj and the interpretation is that of transfer. With aj combined with another verb, the Ns
coming after aj have a different status: those Ns are arguments of the second verb, whereas the preceding
N only is an argument of aj . The interpretation is different in (2b) the workers are not the beneficiaries but
the recipients of the transfer: money 5000 riels go to workers (the aim is to reckon the amount of
money allotted to the labour costs).
The difference between (1a) and (1b) is not as important: in (1a) what is at stake is the everyday
usual process of feeding the dogs, whereas (1b) refers to starving dogs waiting for the food: the sequence on
the right is to a certain extent independent of that on the left.
Comparing those two constructions shows that the demarcation between aj expressing the transfer
and aj as a causative is quite blurred. With the transfer value, aj commutes with cuun (with cuun the
beneficiary has the highest social or parental status). But this local synonymy should not conceal the
semantic differences between the two verbs: with cuun, the focus is on the relation between the two
characters through their social differences. Cuun has not as many uses as aj. Contrary to aj, cuun can
combine with the causative prefix b . This difference can already be found in old Khmer
Benefactive
99
: N
1
V (N
2
)

aj N
3
/ N
1
V (N
2
)

aj t N
3

This second use of aj is closely related to the transfer value. The main difference lies in the
presence of a verb on the left of the sequence: the process referred to by this verb is presented as completed
for the benefit of a character appearing in the sequence on the right. From this point of view, the function of
aj is to point out who benefits by this process.
3) sm ch tlaj aj knm

ask descend cost aj 1sg
Reduce the price for me
4) jk cna tm nuh mk aj ma p

take pot big dem come aj mother part.
Bring me this big pot (the mother to her daughter)
5a) koat t sip aj (cuun) mdaj koat

3sg buy book aj cuun mother 3sg
He bought a book for his mother

99
We take over this terminology from the literature dealing with GIVE-verbs.
126 About aj in Contemporary Khmer

5b) koat t sip aj (cuun) t mdaj koat

3sg buy book aj cuun go mother 3sg
He bought a book, intended for his mother
The benefactive is the second case where aj can commute with cuun. As shown by (5b), aj as
well as cuun can be followed by the verb t go. As for transfer, the presence of t imparts a stronger
autonomy to the sequence coming after aj: buying the book, on the one hand; allocating the book to the
mother on the other hand.
Delegative: N
1
V (N
2
)

aj N
3

The delegative means that someone is making something instead of somebody else: in (6) I drive
the UNESCO managers car instead of him. The presence of a verb in the left sequence must be pointed out;
the sequence on the right is limited to a single N whose relation to the process referred to by the V in the left
sequence is the same as that involving the syntactic subject. The latter is presented as substituting for the
former, which leads to considering one and the same process for the two agents.
6) km bak lan aj niju junnesko

1sg drive car aj manager UNESCO
I drive the UNESCO managers car as his driver
7a) bkpra ttabt nih aj knm p

translate text dem. aj 1sg part.
Translate this text for me!
7b) knm bkpra ttabt nih aj koat

1sg translate text dem. aj 3sg
I translate this text instead of him (delegative) / for him (benefactive)
As shown by the double interpretation made possible by (7b), the benefactive and the delegative
seem pretty similar. To a certain extent at least, for on the other hand, they remain quite distinct: introducing
t after aj is possible only with the benefactive interpretation. The impossibility to have t or any other
V in the right sequence with the delegative can be accounted for by the implicit presence of a V identical to
that in the left sequence, in accordance with the delegative interpretation
1.4. N
1
ku aj V
ku
100
specifies the preceding term as the very best item liable to fill such or such place of argument
of the verb coming after ku. Between ku and the V it is possible to insert aj.
8) bhcnat l nih kmin kanapa na ku aj cap arm sh
election time dem. neg. exist party indef. ku aj
interest at all
Regarding these elections, there isnt one single party worthy of attention
9a) b minp ciwt b kmst mn
older sibling name life older sibling miserable true
ku aj ant nah
ku aj take pity very
Meanup, you have an unhappy life inspiring me compassion
9b) b minp ciwt b kmst mn ku ant nah
older sibling name life older sibling miserable true ku take pity very
Meanup, it is true that you have an unhappy life, one feels compelled to compassion for you (one =

100
ku originates in a verb (suit); it keeps this predicative value in some expressions such as ku pm ku
(ku neg ku) is it all right or not? and ku aj r ku pm aj should we give it to him or not?
Denis Paillard 127

one kind hearted)
The difference between (9a) (presence of aj) and (9b) lies in the fact that the relation between the
N standing as ku scope and the verb is actualized, whereas with ku alone the relation is only potential, the
N being in this case semantically closely related to the V place of argument. This is confirmed by the
impossibility for ku to be used alone in (10): the red complexion of the face has nothing to do a priori with
being fear inspiring; the redness as an argument of fear inspiring actually comes from aj.
10) pel koat k mk koat la krhm ku aj klac
time 3sg be angry face 3sg ascend red ku aj fear
When he gets angry, his face turns red to the point of getting fear inspiring
1.5 Jussive: V

aj

PRED
aj can appear after a V expressing an injunction. The sequence that follows is a predicate
specifying the process expressed by V or one of the arguments. The alternative / aj corresponds to a
difference of interpretation: without aj the term plainly specifies the way the process is carried out; with
aj, the qualification by the postposed predicate is presented as a target.
11a) n sim (sdap ke nijij sn)
remain still (listen people speak first)
Keep still! (listen to the others first)
11b) n aj sim (lec tuk jlj haj)
remain aj still (sink boat at this time PART.)
Keep still! Or youll make the boat sink!
12a) hat j k colct nijij kl c
reason what PART. like speak loud so
But why do you enjoy speaking so loudly!
12b) sm nijij aj kl ci n ban te
part.politeness speak aj loud more deict. can PART.
Could you speak louder?
13a) nijij muj muj ban te
speak one one can PART.
Could you speak slowly?
13b) * nijij aj muj muj ban te
speak aj one one can PART.
In (13b), the term muj muj specifying speak is not a predicate (slowly is the reduplication of the
numeral one), which accounts for the impossibility to have aj :
14a) nijij klj klj t h pel haj
speak short short go finish time PART.
Make it short we have no time left!
14b) nijij aj klj t...
speak aj short go...
To sum up/to conclude...
Should also be noted the difference of interpretation between (14b) and (14a) where the postposed
predicate is reduplicated: the reduplication means that the validation of the predicate depends but on the
interlocutor.

As shown by the series (15), this construction with aj introducing a specification of the process is
not possible with an assertion, a fact which confirms that with aj this qualification is presented as a target.
When aj is replaced by ban (get), the target is presented as actually being reached (cf. (15c)).
15a) rt aj ln
run aj fast
Run fast
15b) koat rt (*aj) ln
3sg run *aj fast
He runs fast (aj is impossible in an assertion)
15c) koat rt ban ln
3sg run acquire fast
He managed to run fast
Finally, it must be noted that this construction can be met in a causative construction (second aj):
16) jk kambt nuh t aj pa samli aj mt t
take knife dem. go aj dad sharpen aj sharp part.
Bring this knife to your father, for him to sharpen it
In this use, it is not possible to have the modal negation km unlike what is to be found with the
permissive (1.6) and the causative (1.7).
Permissive: aj N
2
V / N
1
aj N
2
V (N
3
)
Permissive refers to constructions where aj is followed by a sequence N
2
V (N
3
) where N
2
refers
to a human being and V to a process aimed at by N
2
, but whose achievement depends a priori on the subject
of aj. The latter can be either the locutor (ex. (17a)) or a N or a Pronoun (ex. (17b)). This use is possible in
an injunction as well as in an assertion.
17a) aj km t pa
aj 1sg go also
Let me come with you!
17b) koat aj Dara t
3sg aj Dara go
He let Dara come with him
Causative: aj

N
2
V / N
1
aj

N
2
V
The causative is framed in the same construction as the permissive, the difference lying in the fact
that nothing is said about the relation linking a priori N1 and the process expressed by the V in the sequence
coming after aj.
18a) pel koat mk dl aj koat cam km bntec
time 3sg come arrive aj 3sg wait 1sg a little
When he gets there, tell him to wait for me a little!
18b) koat aj wi cam km bntec
3sg aj 3sg wait 1sg a little
He told him to wait for me for a while
19) nijij aj km sdap pa
tell aj 1sg listen PART.
Tell me about it for me!
Denis Paillard 129

1.8. P1 aj P2: N
1
V

aj

N
2
V (Z) / N
1
V N
2
aj

V (Z)
aj is used as a relator between the propositions P1 and P2. The first clause P1 refers to an event
bringing about the event introduced by the second clause. The verbs liable to appear in P1 are causative
verbs or verba dicendi. It is possible to find in Gorgoniev (1966) and in Bisang (1992) a representative list of
these verbs. It must be noted that for a part of these verbs, a construction without aj is also possible.
Besides, in some cases, the N
2
subject of the verb in P2 can stand before aj. The examples hereafter help to
understand the difference in the interpretation of these three constructions.
20a) km noam koat mk O ml ptah
1sg lead 3sg come O look house
I bring him along to see the house: either 1) for him to see my house and know how it looks like; or
2) for him to know where Im living.
20b) km noam koat mk aj ml ptah
1sg lead 3sg come aj look house
I bring him along to see the house: either 1) the house is for sale; or 2) someone is needed as a
caretaker for the house.
In (20a) (with ) the two events are part of the same whole event associating the two subjects,
whereas in (20b) the two events (therefore the two subjects) are autonomous and independent one from the
other: the V in P2 is aimed as a target involving the N
2
subject of V2, which is confirmed by the impossibility
to have aj in (21) and, conversely that of in (22a):
21) sak pel km noam kn (*aj)
to morrow time 1sg lead child
t sala cam km chit cl le
go school wait 1sg. existence enter play
To morrow, Ill avail on the occasion of taking my son to school to come and see you
22a) km ci nak noam wi aj (*) skoal caowaj kat
1sg. be person lead 3sg. aj know chief province
Im the one who did introduce him to the province governor
22b) km ci nak noam aj wi skoal caowaj kat
1sg. be person lead aj 3sg. know chief province
(he wanted to be introduced to the province governor and) I am the one who managed a meeting
In (22b) where the N
2
subject of V2 comes after aj, the meeting with the governor is presented as a
goal for N
2
, independently of who made it possible for this meeting to take place: P1 is therefore
subordinated to the achievement of P2. A difference similar to that noted between (22a) and (22b) can be
found again considering (23a-b); in (23a), the event expressed by P1 is at stake: what matters is not mainly
knowing where (P2) takes place but rather knowing who made it possible for (P2) to happen (the event
expressed by P2 comes under a negative judgment). On the reverse, in (23b) P1 is totally subordinated to P2:
knowing the location is an appraised information, whereas knowing who made it possible for P2 to be
achieved is a secondary matter.
23a) nak na noam wi aj skoal knla nih
person indef. lead 3sg. aj know place dem.
Who made him know this place? (he wasnt supposed to know this place)
23b) nak na noam aj wi skoal knla nih
person indef. lead aj 3sg. know place dem.
Who made him know this place? (its a good thing)

There being no interdependence between the subjects of the two events when N2 comes after aj is
confirmed by (24) and (25) where P2 refers to a detrimental event. In this case, the N subject of the V in P2
necessarily comes after aj:
24) km somtoh km noam aj b piba dajsa km
1sg. beg forgiveness 1sg lead aj 2sg difficult because 1sg
I beg your pardon for getting you into a difficult situation
25) a twe ac a noam aj ke ml ij dl krusa
2sg. do so 2sg. get aj 3sg look easy get family
What you have done makes our family open to contempt (A father rebukes his son who got arrested
by the police for a theft)
(25) means that the neighbours were only waiting for an occasion to mock at the family
(preconstruction of P2), and its the son who gave such an occasion.
We present two other series confirming the difference of interpretation due to the position of the N
2

in the sequence:
26) A mother is complaining about his son who doesnt want to study:
26a) a muj n ba km mn bkm wi
anaph. one deict. if 1sg. neg. force 3sg.
aj rin te wi at rin te
aj study part. 3sg. neg. study part.
This one, if I dont force him to study, he wont do it (he refuses to study, he has to be made to do it)
26b) a muj n ba km mn bkm aj
anaph. one deict. if 1sg. neg. force aj
wi rin te wi at rin te
3sg. study part. 3sg. neg. study part.
This one, if I dont force him to study, he wont do it (he will never consider doing it on his own
initiative, he has to be prompted to do it)
In (26a), the presence of the subject of P2 preceding aj emphasizes the interdependence between
mother and son as regards studying, whereas in (26b) (subject of P2 coming after aj) studying is
presented as an aim the mother endeavours to reach. It can also be observed that aj can hardly be removed.
27a) caowaj srj a mn ban bkoap aj a
boss woman 2sg neg. get order aj 2sg
twe te a twe klun a ta
do part. 2sg do body 2sg restr.
Your boss did not tell you to do it, you did it on your own ?!
27b) caowaj srj a mn ban bkoap a aj
boss woman 2sg neg. get order 2sg aj
twe te koat aj nah pse twe ta
do part. 3sg aj person else do part.
Your boss did not ask you to do it, she asked somebody else to do it
27c) cawaj srj knm bkoap mdaj knm aj
boss woman 1sg order mother 1sg aj
knm rk lj aj koat
1sg seek money aj 3sg
My boss ordered my mother to ask me to go and fetch money for her (the boss)
Denis Paillard 131

In (27a) and (27b), the verb bkoap order (P1) comes under a negation: with the anteposition of
N
2
(ex. 27a), the independence of P2 is reinforced: P1 is not the event which brings P2 about. With the
postposition of N
2
(ex. 27b), what comes under the negation is the identity of the agent of P2. Finally, in (27c)
where we find both a N object of bkoap order and a postposed N, the N object takes over from the agent
of bkoap as regards the validation of the process referred to by P2.
When the V of P1 is a strict causative verb, such as bndl cause, N
2
postposition is only possible:
owing to its semantic value, the verb bndl cannot express a relation between the two subjects:
28) koat kmin tw j dal min prjaj dl krmhn te
3SG. NEG. do INDEF. REL. have interest reach Firm PART.
koat bndl aj ta krmhn kat lj n
3SG. cause aj only firm lose Money DEICT.
He doesnt do anything useful for the firm, all he does is getting it into losses!
We now present a last series with tw (do) as a causative verb and with an inanimate N as the
subject of the V of P2.
29a) tw lan aj (*) ch sn cam t
make car aj * work first wait go
Fix the car before you leave! (manage to make it work before leaving)
29b) nak na tw lan aj (*) ch
person indf. make car aj * work
Who fixed the car? (I thought there was no one to fix it)
29c) nak na tw aj (*) lan ch
person indf. make aj * car work
Who made the car work? (I dont want it to be put in working order)
29d) pel km tw (*aj) lan ch haj
time 1sg. make *aj car work part.
km k ripcam dam baj
1sg. then prepare cook rice
After getting the car fixed, I cooked the rice
In (29), the variation (with an equal lexicon) involves a. the place of the subject of P2 (lan car)
before or after aj; b. the presence or not of aj between P1 and P2. In (29a) and (29b), lan must
necessarily stand before aj: P1 is first. In (29c) lan must necessarily stand after aj: stating that the car is
working (P2) is what leads to wondering who is responsible for fixing it (cf. the comments given on the
above 22b et 23b examples). The impossibility to have aj in (29d) is due to the fact that the event fix the
car is part of a succession of events given in order and therefore taken as a whole. In other words, entering a
succession makes it impossible for P2 to be dissociated from P1
The construction P1 aj P2 shows various degrees of autonomy of P2 as regards P1, depending on
the position of the subject of the V in P2. When N2 stands before aj, the event expressed by P1 is
prominent (the autonomy of P2 is quite relative). When N2 comes after aj, P2 is presented as a goal to be
reached (or that has been reached), and P1 is subordinated to the achievment of this goal.
1.9. Optative: aj+ta p and km+aj+ta p
In this case aj is at the initial position
101
and is followed by ta (only). aj combined with ta, as
a marker of restriction, means that p is the only thing taken into account as regards the evaluation /
interpretation of a situation contextually introduced. This situation takes into account a set of possible options,

101
It can be preceded by the modal negation km; cf. the hereafter examples (33) and (34).

and the selection of p as the only one actually selected is presented as kept away from any kind of control,
whether it be virtual or actualized. This accounts for the fact that nothing can be found on the left of aj, not
even a subject / agent. Given all that can be associated with this set, the only thing deemed relevant is p,
whether actualized or not.
30) wi skct tw tam a taah aj ta a nijij l n wi
3sg. agree do follow 2sg. all aj ta 2sg. speak well with 3sg.
He agrees to do anything you want, on the only condition that you speak kindly to him
31) aj ta ban t aj km tw j k km tw da
aj ta acquire go aj 1sg. do what part. 1sg. do also
On the only condition that Im free to go away, I can be asked anything, I will not refuse to fulfil it
32) kme n srala w wi aj ta wi
child Deict. love father 3sg. aj ta 3sg.
k w wi wi m plim
see Father 3sg. 3sg. smile immediately
This child does love his father: catching sight of his father is enough for him to smile
In (32), what is selected is the complex sequence formed by see his father smile right after.
Examples of aj ta with the modal negation km:
33) wi skct tw tam a taah km
3sg. agree do following 2sg. everything km
aj ta a nijij akr n wi
aj ta 2sg. speak nastily with 3sg.
He agrees to do anything you want on the only condition that you do not speak to him nastily
34) wi kranta jk a t km mn j te km aj
3sg. only take 2sg. go put in jail neg. indf part. km aj
ta wi waj tamtit nh tk ci smna t
ta 3sg. beat addition deict. consider be luck part.
He (the cop) put you into police custody, no use making a fuss about it, the important thing is that he
didnt rough you up, you should consider yourself lucky!
In (33) and (34) km indicates that neg + p (not speak badly) is the only important thing.
Lets point out that in some cases, either aj ta or km aj ta can be found:
34a) kme n mn d ci jamec wi aj
child deict neg know be how 3sg aj
ta wi k w wi wi jm plim
ta 3sg see father 3sg 3sg cry immediately
I dont know what is the matter with this child: as soon as he sees his father, he starts crying
34b) kme n mn d ci jamec wi km aj
child deict neg know be how 3sg km aj
ta wi k w wi wi jm plim
ta 3sg see father 3sg 3sg cry immediately
I dont know what is the matter with this child: seeing his father is enough to start crying right away
With aj ta, considering see his father - cry is nothing but a mere statement, grounding the
remark I dont know what is the matter with this child. The presence of km means that the reference value
is not see his father - start crying, which underlines the contradiction between see his father and start
Denis Paillard 133

crying. This is confirmed by the fact that in (32) only aj ta is possible, km aj ta being impossible (the
sequence which makes the scope for aj ta is appraised).
1.10. ta aj and sdj aj: criticize, rebuke: [N
2
V] N
1
ta aj / sdj aj N
2

Combined with two verba dicendi ta and sdj, aj mean criticize (ta), rebuke (sdj). This
interpretation is framed by the following rules a. the N coming after aj refers to the person being criticized
or rebuked; b. the sequence expressing the event giving rise to the criticism or the rebuke is present in the
left context, but resuming it is impossible after aj.
35) mn n ta (*sdj) aj ke ml klun a sn
before part. ta *sdj aj people look oneself 2sg first
Before criticizing the others, just look at yourself! (Someone criticizes the people around him about
what they wear)
36a) ba wi mk jt km ta aj wi
if 3sg come late neg.mod. ta aj 3sg
If hes late, dont be critical!
36b) ba wi mk jt km sdj aj wi
if 3sg come late neg.mod. sdj aj 3sg
If hes late, dont rebuke him!
37) mother asks her husband not to rebuke their son for being out every night
37a) pel wi mk km sdj aj wi cam
time 3sg. come neg.mod. sdj aj 3sg. wait
sak nijij pnjl wi tam smruj t
tomorrow speak explain 3sg. following quiet go
When he comes back, dont shout at him ! Wait until tomorrow to have a quiet explanation with him!
37b) pel wi mk km ta aj wi cam
time 3sg. come neg.mod. ta aj 3sg. wait
ak nijij pnjl wi tam smruj t
tomorrow speak explain 3sg. following quiet go
When he comes back, dont criticize him! Wait until tomorrow to have a quiet explanation with him!
38) ba wi mk jt km ta (*sdj) j aj wi
if 3sg come late neg.mod. ta *sdj indef aj 3sg
If hes late, dont tell him anything!
39) a sdj (*ta) aj koat r j pt t
2sg. sdj ta aj 3sg. matter indef. be true go
What are you rebuking him about exactly?
The differences of interpretation between ta aj and sdj aj can be grounded on the semantic
core of the two verbs: ta deals with words and ways of saying things, whereas sdj deals with events, as
shown by a comparison between (38) and (39): in (38) the indefinite j refers to the class of possible words,
and ta only is possible; whereas in (39) r story refers to an event (which has to be identified) and sdj
only is possible
102
.
The negative dimension of the verbs (criticize, rebuke, shout at) can be explained by the fact
that the implication of N
2
in an event giving rise to a negative (verbal) reaction (be late, be out every

102
This difference has to be backed out by a systematic study of these two verbs

night) gets N
1
to say something about it: a reproachable behaviour leads to N
1
seizing verbal power on N
2
:
he is putting him on (personal) trial.
2. A synthesis of the ten classes of uses
This presentation of ten large classes of uses of aj has made it possible to show that each use can
be characterized by a set of specific syntactic properties grounding the interpretation of the utterance with
aj. This presentation has also made it possible to bring out a characteristic feature at work in all the uses of
aj : the partial autonomy of the sequence coming after aj. As a rule, this autonomy of the sequence
following aj comes together with the - possible or necessary - presence in this sequence of a V different
from aj . Owing to this verb, the sequence can refer to an event which is part of the complex event
expressed by the whole utterance. As regards the presence of a V in the sequence before aj, it should be
noted that this is the case with the following uses: benefactive, delegative, jussive, P1 aj P2 and
criticize/rebuke.
We hereafter resume the ten types of uses in reference to the autonomy of the sequence coming after
aj.
Transfer is characterized by the possibility to introduce a verb after aj. The presence of a V (normally t
to go), which is not compulsory, means that the transfer is not considered as the mere passage of an entity
from a subject S1 to a subject S2 (S1 being the active part in this passage). With a postposed V, putting this
entity in relation with S2 is considered as prior, and taking S1 into account is subordinated to this relation.
Benefactive comes with a V standing both on the left and of the right of aj, The one on the right is not
compulsory contrary to the transfer case. Benefactive means that the process referred to by the verb standing
before aj is not considered only from the point of view of the subject agent (S1) of this process, but from
the point of view of another subject S2 standing on the right of aj. The presence of t on the right of aj
reinforces the subordination of the process achieved by S1 to the interest of S2.
Permissive and causative: the compulsory presence of a verb after aj means that the subject-agent of aj
is taken into account as it allows the event referred to by the V to be possible or achieved.
P1 aj P2 is characterized by the compulsory presence of a V both on the left (P1) and on the right (P2) of
aj. The V of P1 has a causative value and is understood as the achievement of the event referred to by the
V in P2. However, when the subject of the V in P2 comes back to the left of aj, the event referred to by P1
gets its own autonomy as an event.
Jussive and ku: alternative / aj. The mere possibility of a construction without aj means that the event
is referred to mainly by the verb other than aj; This V is on the left in the jussive case, on the right in the
ku case. In the jussive case, the predicate on the right of aj is not interpreted as a mere determination of
the process, but as a goal to be reached - the validation of the process being considered regarding the
achievement of this goal.
Delegative: the impossibility to have a V in the sequence coming after aj and the compulsory presence of a
V before does not make an exception of the delegative. As shown above, the validation of V by a subject S1
is not at stake for itself: it matters only in regard with the substitution of S1 to a subject S2 standing on the
right of aj. The delegative implies a verb, but this verb is not explicit, being the same as in the left
sequence, or already present in the left context.
Optative: no sequence at all before aj : the only thing which matters is the (actual or aimed at)
achievement of the only event referred to by the sequence coming after aj.
Criticize rebuke: the impossibility to have a V coming after aj lies on a mechanism comparable to that
described in the delegative case, but for one difference: the V which is not explicited on the right of aj is
present in the left context and not directly in the sequence before aj. But the V has not the same
interpretation in the two positions. In the left context, it refers to an event (which is considered as having a
negative value); in the utterance with aj preceded by the verbs ta and sdj, it corresponds to something
Denis Paillard 135

that is said (meaning a predication): the agent of the process on the left is resumed by the N coming after aj
as a subject of whom is said that he is responsible for a process.
3. Characterization of aj.
As indicated before, it is not possible to account for these ten uses of aj basing on a central value
called transfer. It appears that aj is not just a verb like any other, but justifies its definition as a
metapredicate, with the following characterization: its function is to put in relation two events E1 and E2,
the first one being introduced as the trigger of the second one:
E1 aj E2
The relation thus established between E1 and E2 gives aj a causative dimension, but this
relation should not be reduced to a mere causal relation. In 2. the autonomy of the sequence coming after aj
has been pointed out; in the above notation, this sequence corresponds to E2. This autonomy of E2 leads to
the assumption that E2 comes first, ant that E1 is only taken as the trigger of E2. This primacy of E2 means
that E2 is introduced independently of E1, even though the realization of E2 interacts with that of E1.
An event can be minimally defined as involving a subject (S) and a predicate (p). Our hypothesis on
the semantic core of aj can therefore be schematized as follows:
S
1
p
1
(E1) aj
103
S
2
p
2
(E2).
This semantic function is at work in all the uses of aj. All these uses correspond to various modes
of realization coming from the specific units embodying the sequences S
1
p
1
and S
2
p
2
. The sequence
corresponding to the use P1 aj P2 gives the largest extension: S
1
p
1
and S
2
p
2
correspond each to a
clause formed on the pattern N V (XY). The transfer value corresponds to the minimal extension, aj
being the only verb, S
1
p
1
and S
2
p
2
reducing to Ns.

From this viewpoint, describing such or such use of aj consists in interpreting the different
constituents of the utterance as framed by the S
1
p
1
aj S
2
p
2
pattern. We hereafter present a table showing
the organization of the ten uses of aj with an indication of the element of the abstract pattern realized by
each constituent
104
. When an element of the abstract pattern is not realized, we use the symbol. An element
of the abstract pattern can be realized by more than one constituent. When the materialization of an element
of the abstract pattern is optional, we put the corresponding element into brackets.
Uses S
1
p
1
aj S
2
P
2

Transfer N
N

aj
aj
N
N
N
V + N
Benefactive N
N
V + N
V + N
aj
aj

N
V + N
Delegative N V + (N) aj N
ku N ku aj (N) V
Jussive the addressee V aj PRED
Permissive (N) aj N V
Causative (N) aj N V
P1 aj P2 N
N
V
V + N (= S
2
)
aj aj N

V
V

103
Keeping aj in the notation of its semantic characterization comes from the fact that it works as what we
called a metapredicate: it does not express an event as such, but plays a central role in the complex event
corresponding to the relation established between S
1
p
1
(E1) and S
2
p
2
(E2).
104
In this table, we only mention the constituents in direct relation with the abstract pattern S
1
p
1
(E1) aj S
2

p
2
(E2).

Optative aj N V
Criticize /
Rebuke
N
N
ta
sdj
aj N
N

Conclusion
This approach of aj leads to stating that the various uses and values of this verb are always
constructional. It makes it impossible to consider one of the values (that of transfer in the present case) as
more basic than the others. It breaks the widely spread idea according to which the lexical units encodes
entities or events of the world (a central hypothesis in Newmans works on GIVE). It brings forward the
unity and coherence of aj in its various uses.
The semantic identity of aj is to be found in everyone of its uses, through variations coming from
the other constituents of the utterance. The characterization we have put forward appears as a schematic form.
This means that the interaction between aj and some of the items of the context is double: a. as a scheme, it
organizes the elements of the context, framing them in a given pattern; b. as a form, it gets its substantial
value (its content) from the lexical units embodying this abstract form in a given construction.
Abbreviation
DEICT deictic or demonstrative
RELAT relative
INDEF indefinite
PART particle
NEG negation
1SG 1
st
person of singular personal pronoun
2SG 2
st
3SG 3
st
References
Benveniste, E. (1993) Vocabulaire des Institutions Indo-Europennes 1, Paris: Editions de Minuit.
Bisang, W. (1992). Das Verb im Chinesischem, Hmong, Vietnamesischen, Thai und Khmer. Tbingen:Gunter
Narr Verlag.
Gorgoniev, Iou (1966) Grammatika khmerskogo jazyka, Moskva
Heine B. & Kuteva T. (2002) World lexicon of grammaticalization, Cambridge University Press.
Jenny, M. (2005), The verb system of Mon, Arbeiten des Seminars fr Allgemeine Sprachwissenschaft, Nr.
19, Universitt Zrich.
Khin S. (2002) La grammaire du khmer moderne. Paris: Editions You Feng.
Newman, J. (1996) Give: a cognitive linguistic study, De Gruyter
Newman, J. (ed.) (1997) The linguistics of giving. Studies in Typological Linguistics, 36, John Benjamins.
Oguma, K. (2006) Identit et variation du verbe yaru en japonais, in: Constructions verbales et production
de sens, Besanon: Presses de lUniversit de Franche Comt, pp. 307 321.
Paris, M. C. (1998) Syntaxe et smantique de quatre marqueurs de transitivit en chinois standard: ba, bei,
jiao et rang, La transitivit, Lille, Presses universitaires du Septentrion, Coll. UL3, p. 355-370.
Rangkupan, S. (2005) The Syntax and Semantics of GIVE Complex Constructions in Tha, Paper
presented at the 2005 International Course and Conference on Role and Reference Grammar,
Institute of Linguistics, Academia Sinica, Taipeh, Tawan.
Thepkhanjana, K. & Uehara, S. (2008), The verb of giving in Tha and Mandarin Chinese as a case of
polysemy: a comparative study, Language Sciences 30, pp. 621 651.
Yap & Iwasaki (1998) A typological and semantic analysis of the relationship between causatives and
passives in the periphrastic give construction of some East and Southeast Asian languages. Paper
presented at the Eighth annual meeting of the Southeast Asian Linguistics Society.
Denis Paillard 137

Yap & Iwasaki (2003) From causatives to passives: A passage in some East and Southeast Asian languages,
Cognitive Linguistics and Non-Indo-European languages, Ed. by Casad & Palmer, Mouton de
Gruyter, Berlin New York, p. 419 445.

Suwilai Premsrirat. 2011. Pearic, a Dying Branch of Austroasiatic Languages and Its Struggle for Survival. In Sophana
Srichampa and Paul Sidwell (eds.) Austroasiatic Studies: papers from ICAAL4. Mon-Khmer Studies Journal Special
Issue No. 2. Dallas, SIL International; Salaya, Mahidol University; Canberra, Pacific Linguistics. pp.138-153.
138
Pearic, a Dying Branch of Austroasiatic Languages and Its Struggle for Survival
Suwilai Premsrirat
Center for Documentation and Revitalization of Endangered Languages
Research Institute of Languages and Cultures of Asia (RILCA)
Mahidol University, Thailand
1. Pearic languages in the Austroasiatic Language Family
Austroasiatic languages are the substratum of indigenous languages in mainland Southeast Asia. 160
Austroasiatic languages can be found in the large area of mainland Southeast Asia. According to Diffloth
(1974), Austroasiatic languages are divided into two main sub-families; Munda and Mon-Khmer. The latter
consists of Northern, Southern and Eastern Mon-Khmer, of which Pearic is a branch. Other branches in
Eastern Mon-Khmer are the Khmeric, Bahnaric and Katuic.
Pearic belongs to the Eastern Mon-Khmer division of the Austroasiatic language family. There are
seven languages in the Pearic branch: Chong, Chung, Kasong, Samre, Suoi, Somray and Pear. Some are
found along the Eastern border of Thailand and across the border to Cambodia. Even though there are record
numbers of Pearic languages and speakers in various locations throughout Cambodia (as shown in Map1 and
Map2), they are not easily found these days.

Map 1: Pearic location in Eastern Thailand and Cambodia (Isara Choosri, 2007)
Pearic a Dying Branch of Austroasiatic 139

2. Pearic languages in Thailand
Of the more than 70 languages of Thailand, 23 are Austroasiatic languages, and among those are
four small Pearic languages: Chong, Kasong, Samre, and Chung (Suwilai et al. 2004). They are found
mainly in the Eastern region near the Cambodian border, except for the Chung people who are found in
Karnchanaburi in Western Thailand, where their ancestors were brought as prisoners of war in A.D.1830.
2.1 Pearic languages as seriously endangered languages
The twenty first century is an age of rapid change, and bio-cultural diversity is being threatened.
Ninety percent of world languages are estimated to face extinction by the end of this century, and these are
mainly ethnic minority languages (Krauss 1992). For the indigenous Austroasiatic Language Family, the
whole branch of Pearic languages is seriously endangered (Suwilai 2007). The changes in the ecology of
language are caused by the global economy, global culture, and global communication. Powerful mass media
channels are streamed right into the home, even in remote areas and they mainly use dominant languages.
This is exacerbated by policies in national language and education that only promote the use of the
national/official language for formal occasions, as the medium of instruction in schools, and as the language
of the mass media. Also, increased options for job prospects for the young require the use of a widely-used
language. As such, the younger generation does not see the value of, and even has a negative attitude toward
its own ethnic languages. In turn, ethnic languages become stigmatized. As young people speak the official
language more and more, they speak their ethnic languages less and less, or even stopped speaking them
altogether. At the moment, in Thailand there are 15 seriously endangered languages; nine are Austroasiatic
languages, and four of them are Pearic languages. Three Pearic languages, such as Samre, Kasong and Chung
(Sa-oc), are at the last stage of endangerment prior to extinction, whereas Chong is at the second from the last
stage (Suwilai, 2007).
According to Fishmans GIDS, all languages of Thailand are on the weak side. They are at different
stages of endangerment; Kasong, Samre and Chung (Sa-Oc) are at the last stage of endangerment before
extinction, because there are only a few fluent speakers remaining. Chong, Lawa (Gong), and So (Thavung)
are at stage 7 because only the older generation still uses the language enthusiastically among themselves, but
not the younger generation. The languages in stage 6 are Urak Lawoi, Moklen / Moken, Mpi and Bisu which
are used in the home and community, but not enthusiastically. Mlabri and Maniq (Sakai) are still used in
daily life, but the number of speakers is small. There are not more than 300 speakers for each of these groups.
Lavua and Nyah Kur are larger groups, but the children still use their ethnic language in some communities.
However, since Lavua, Nyah Kur, Chong and So (Thavung) are now undergoing a language revitalization
program, and are being taught as subjects in schools in the area, they may be placed at stage 5.
Stage 8 So few fluent speakers that community needs to re-establish language norms;
often requires outside experts (e.g.,linguists). [Kasong,Samre and Chung (Sa-
oc)]
Stage 7 Older generation uses language enthusiastically but children are not learning
it. [Chong, Lawa (Gong),So (Thavung)]
Stage 6 Language and identity socialization of children takes place in home and
community., [Urak Lawoi, Moklen, Mlabri, Maniq (Sakai), Mpi and Bisu]
Weak Side
Stage 5 Language socialization involves extensive literacy, usually including non-
formal L1 schooling or teaching L1 as a subject Local Studies in school
[Chong, Lavua, Nyahkur, and So (Thavung)].
Stage 4 L1 used in childrens formal education in conjunction with national or
official language.
Stage 3 L1 used in workplaces of larger society, beyond normal L1 boundaries.
Stage 2 Lower governmental services and local mass media are open to L1.
Strong Side
Stage 1 L1 used at upper governmental level.
Table 1: Eight stages of language endangerment according to Fishmans GIDS
(Adapted from Suwilai Premsrirat, 2007)
140 Suwilai Premsrirat

The Pearic speaking people, especially, the Chong, Kasong, and Samre, are believed to be the
indigenous people of eastern Thailand and the adjacent area in Cambodia, which was a part of the Ancient
Khmer Empire. At the moment, the majority of the Chong are found in the Khao Khitchakut district and
some are in the Pongnamron district, Chanthaburi. Although Chong descendants are found in a large area in
Chanthaburi, at the moment fewer than 2,000 people still speak some Chong, and the ability among people of
different age groups varies widely. Though Elderly people still speak the language among themselves, they
have been heavily influenced by Thai vocabulary. Most Chong descendants under thirty years of age do not
speak Chong; Thai is their first language. Only young Chong people can speak the language. Good Chong
speakers may not number more than 200 individuals. They are found in Khao Khitchakut area in Chanthaburi,
only 3 elderly speakers are found in Pongnamron, Chanthaburi. Chong is therefore classified as a language at
stage 7 of Fishmans GIDS for threatened languages, where the language is used by the older but not younger
generation. A language at this stage needs to multiply the use of the language in the younger generation
(Fishman 1991).
As for Kasong, Samre and Chung, the situation is worse. Kasong speakers are found in only three
villages (Khloangsaeng, Danchumphon, and Padao). Even though there was a record of more than 3,000
Kasong people (Chong people of Trat) throughout the Borai subdistrict, Trat province; there are now not
more than 50 Kasong speakers and not more than ten good speakers. The Samre are found in two villages,
with about 30 speakers and only a few good speakers. It is expected to become extinct in the near future
(Pornsawan 2001). Samre was also recorded as spoken in Sanamchaikhet, in Chachoengsaw province. Samre
descendants remain there, but from my last trip to this area more than ten years ago, the people had already
lost their ethnic language. They could recall only few words. The abbot in a Buddhist temple said that he
used to hear people speak the language when he was young, but had not heard it for a long time. There might
be some Samre speakers in the Cambodian forest, but no investigation has yet been undertaken. Both Kasong
and Samre are known by Thai people as Chong. Chung (Sa-oc) is closely related to Chong. Sa-oc skin
disease is a Cambodian Khmer exonym (Diffloth pers. Comm.) used for referring to these people in their
ancient empire. This name is known in the area as Auut. Fewer than 50 speakers are found in Srisawat
district, Kanchanaburi province. Only a few good speakers can be found. Their ancestors were prisoners of
war in the early Bangkok period, but most Chung descendants now speak mostly Thai. Some who live in the
nearby area can speak Khmu. A village of Chung-speaking people has been found in Kampong Som in
Cambodia (Isara 2009).
Samre, Kasong, and Chung (Sa-oc) can also be classified at stage 8 of Fishmans GIDS, where there
is a social isolation of the few remaining speakers from the minority language. A language at this stage needs
to be recorded for possible later reconstruction. Pornsawan (2001) wrote a doctoral dissertation on Samre
grammar, where Samre phonology and syntax are documented. As for the Kasong, three M.A. students in the
Linguistics Department at Mahidol University worked on Kasong phonology (Noppawan 2003), Kasong
syntax (Sunee 2002), and the Kasongs language attitudes (Suwapat 2003). Even though these language
groups are also found in Cambodia, according to Gerard Diffloth (pers. comm.) the whole branch of Pearic
languages is endangered.
2.2 Phonological, lexical and syntactical characteristics of Pearic as a group of endangered languages
(1) The shift of the Pearic lexicon toward the dominant language in the area is quite common. A
large percentage of words are Thai loan words. There are also Khmer loan words in case of Chung. Even
though some of the basic Pearic vocabularies are still retained, it is obvious that loan words from the
dominant languages are prevalent and spread beyond nouns and verbs into all parts of the lexicon, including
closed classes of grammatical words. More than half of the lexical inventory is influenced by the dominant
language. Therefore it is now urgenct for lexical documentation and text materials or discourse collection, so
that knowledge systems and local wisdom inherited in the Pearic languages can be preserved as much as
possible before they are all lost forever. The following examples demonstrate the basic Pearic vocabulary that
is still retained.

CHONG CHUNG KASONG SAMRE
Number
One mo
j mj mj mj
Two phj pr p p
Three phw phw ph ph
Four phn phn phn pun
Kinship Terms
Mother me me mi mim
Father u u ku ku
Husband kal klo kl klu
Offspring (children) ken ken ken kin
Natural phenomena
Water tak tk tak tak
Fire pew pliw plew pliw
Mountain kn no n nu
Rain kama km kam kam
Sun ti ti tani sani
Table 2: Pearic basic vocabularies still retained
Pearic basic vocabularies still retain numbers, kinship, natural phenomena, plants, animals, basic
verbs and nouns. Some examples are shown in Figure 3. However, the examples of Chong sentences below
illustrate that there are heavy influences from Thai. Even basic words such as nouns, verbs, pronouns,
grammatical words, etc. have been borrowed from Thai (Thai loan words are in bold)
ban lu ha tk
R1
paj
R2
se
R1

house Uncle- Hang sells twenty)
LungHangs shop sells (it) for twenty baht.
k
R3
naj mnam in
R1
raj
R3
tu
crocodile in river have ten clas.
There are ten crocodiles in the river.
pj
R3
calat kwa dak
R1
na
it clever than they fp.
It is smarter than the others.
mu pj
R1
cp wiwat sa
R1

they like argue each other
They like to argue with each other.
pn kakaj m
R1
sa
R1

friend trade together
They do business together.

mu pj
R1
si ha kon
group it four five class.
Four or five of them.

(2) There have been great changes in syntactic structure. Grammatical words, auxiliaries, final
particles and conjunctions are heavily borrowed from Thai. Only a few affixations are found and none are
productive. The syntax of all Pearic languages is becoming more like Thai though some Mon-Khmer
characteristics can still be observed.
pj
R1
ih
R1
hoc
R1
ih
R1

it not die not
It does not die.
2.1) The negative construction (Neg) -Verb-Neg in Chong and Chung is still retained, though the
gradual change to the Thai negative construction can also be observed. The last example sentence below has
the same negative pattern as that in Thai.
kah
R1
nas ih
R1
j
R1
i
R1
t cew
R1
ih
R1

know book not fp. I must go not
(I am) illiterate. I must not go.
kit ih
R1
k
think not out)
(He) cannot think of (a way to do).
Examples above illustrate negative constructions in Chong and Chung that are different from Thai
but similar to Khmer. This confirms the settlement of Pearic people with the Khmer and probably dates from
the ancient Khmer empire.
2.2) The Noun Phrase / Noun compound in Kasong and Chong illustrates the original construction
and the alternative construction influenced by Thai.
KASONG
English Gloss
Original Alternative Thai
mt khli khli mt ph n
Younger and elder sibling
younger elder (sibling) elder younger (sibling) elder younger (sibling)
san kl kl san phu mi
Husband and wife
wife husband husband wife husband wife
san khen khen san luk mi
Wife and children
wife child child wife child wife
CHONG
English Gloss
Original Alternative Thai
mt
R3
aj
R3
aj
R3
mt
R3
n ta
Face and eyes
eye face face eye face eye
bot
R1
l
R1
l
R1
bot
R1
pi n
Younger and older sibling
younger elder (sibling) elder younger (sibling) elder younger (sibling)
me
R3
u
R1
u
R1
me
R3
p m
Mother and father
mother father father mother father mother
Table 3: Noun phrase / Noun compound constructions in Kasong and Chong

2.3 Affixes in Kasong and Chong
Chong locative and causative prefixes and instrumental infix are found, but none are productive.
CHONG
Affixes Function Example
d
R1
on pad
R1
above
mun
R1
behind pamun
R1
behind
re
R1
in pare
R1
inside
/pa-/ Locative
tw
R3
other patw
R3
elsewhere
/ma-/ Causative hoc
R1
die mahoc
R1
to kill

KASONG
Affixes Function Example
/-n-/ Instrumental
khet
to comb
khanet
comb

kew
to harvest
khanew
sickle

pk
to wrap
pank
package

kk
to carry on the shoulder
kank
shoulder pole
Table 4: Non-productive Kasong and Chong affixes
(3) The variation and change in phonetics and phonology of Pearic languages as a result of influence
from Thai is obvious. There is a great variation in the pronunciation of elderly and younger speakers in both
segmental and supra segmental phonemes. There has been a trend that the young are losing not only their
ethnic vocabulary, but also outstanding contrastive registers. Some still show the contrast, but very weakly.
Register complex and tonogenesis found in Pearic languages at the moment have been obviously influenced
by the dominant Thai. Chong and Chung are considered register languages, whereas Samre is considered as a
tonal language and Kasong is in transition with both 2 contrastive registers and 2 contrastive tones. Examples
are shown below.
R1 (Mid clear)
R2 (High creaky / Glottal
constriction)
R3 (Low
breathy)
R4 (Low breathy followed by high
glottal constriction)
ktak peanut ktak tongue ta k water
Chong
kl bone
kl

kl
husband, male
kl
temporary wood bridge
tak bean,
peanut
tak tongue ta k water
Chung
ml eel ml salty
R1 (mid) R2 (high-falling)
R3 (operated
low)
R4 (operated, mid, high falling)
tak big ktk tongue t k water t k wet
kl bone
kl
husband

k k green frog k long
lale rotten
worm
p three p water p watch
Kasong
c dog c sour


Mid Low Mid-low
klu bone klu husband klu log --
puc
( )
kapuc turn upside down
puc to dip
water
--
pat to lick pat to slice
pt to walk
pass
--
Samre
suto dance su to smell su to tell --
Table 5: Register Complex and Tonogenesis in Pearic languages
The changes to and decline of the Pearic languages is obvious. But what is being done? Linguists
have been documenting and describing various aspects of the Pearic languages as much as possible as
recorded in Figure 8 below.
Survey Pearic languages
and Stage of endangerment
CHONG
CHUNG
(Sa-oc)
KASONG SAMRE
Documentation and
Description
Stage 7 Stage 8
Sociolinguistic/ Attitude Survey or Language
vitality
-- --
- Phonology
- Register/ Tone/ In Transition
- Syntax
- Lexicon/ Dictionary/ Wordlist
- Dialects -- --
- Songs / Poems -- -- --
- Folktales
- Orthography -- --
Table 6: Pearic Language Documentation and Description
Apart from language documentation and description carried out by linguists, the language speakers
themselves can also document their own language and culture from the needed perspective of native speakers
of the language. The Lexicon/ Dictionary Compiling is urgent for all endangered languages. It can be done
by both linguists and language speakers. The first Chong dictionary was carried out by the community in
2002; though it is more like a list of words and phrases. The words were mainly collected by the Chong from
Khlong Phluu village. Then in 2009, linguists published a Chong-Thai-English Dictionary containing three
dialects. The two Chong dialects, Northern and Southern Chong, are from the Khao Khitchakut area and one
dialect from Pong Namron (Eastern Chong dialect).
The Chong still have enthusiastic groups of people who want to keep their language alive, especially
people of middle age who are active and see the value of their own mother language; they actively participate
in the Chong revitalization program. However, other Pearic languages such as the Kasong, Samre and Chung
do not have such enthusiasm. The Kasong and Samre have only a few elderly speakers. They are found in the
border area of Trat province. It is not likely that these languages will survive following the death of the

remaining elderly speakers. This prognosis can also be applied to the Chung which are in Srisaswat district,
Karnchanaburi province in Western Thailand.
3. Pearic speakerss struggle for the survival of their language
Apart from the documentation and description of Pearic languages by linguists, there are the
reactions from grassroots communities for language revitalization programs with guided cooperation from
linguists. Language Revitalization Programs are an attempt to add new linguistic forms or social functions to
embattled minority languages with the aim of increasing the languages uses or users (King, 2001, p. 23)
According to Crystals Six Postulates of Language Revitalization, an endangered language will
progress if its speakers: 1) increase their prestige in the dominant community; 2) increase their wealth
relative to the dominant community; 3) increase their legitimate power in the eyes of the dominant
community; 4) have a strong presence in the educational system; 5) can write their language down; and 6)
can make use of electronic technology. The Chong comprise the first endangered linguistic group that has
undergone the language revitalization program.
The Chong Language Revitalization Program (CLRP) began with cooperation between Chong
elders and Mahidol linguists with strong community motivation and commitment. Financial support was
received from the Thailand Research Fund (TRF) as community-based research conducted by the speakers
themselves. The technical support as well as psychological and emotional support was received from the
Research Institute for Languages and Cultures of Asia (formerly Institute of Language and Culture for Rural
Development (Mahidol University-Linguists and education experts).
The Chong Language Revitalization Program (CLRP) is composed of orthography development,
literature production, curriculum development, teaching Chong as a subject in school (by native language
speakers), and a Chong community learning center for public at large.
The orthography development component is a complex process of developing a writing system for a
previously an unwritten language. Native speakers have to be actively involved in the orthography
development process with support from the linguists. The process involves selection of a script writing
system or alphabet for the standardization of the systems and vocabulary expansion to produce literature and
reading materials.
3.1 Chong language revitalization program (CLRP), the first attempt
The Chong Orthography was based on linguistic research. The Thai script has been selected for use.
Three criteria are considered Linguistic factors (simple, phonological adequacy) socio- psycholinguistic
factors, and technical factors. Orthography is an important tool for recording the Chong language and local
knowledge, and the writing system is used for teaching younger generation and as a symbol of ethnic identity.
Community involvement is indispensable to the success of the project. The newly developed
orthography has to be accepted by the community. The community should actively participate in the
development process from the very beginning; selecting the script, looking for the different or outstanding
features of the target language, looking for minimal pairs, selecting the symbol for each sound, looking for
examples and consonant / vowel / tone or register. After that the tentative orthography has to be tested to see
how readable and acceptable it is to the community. Then, an alphabet chart with keywords and pictures is
able to be produced. The Chong people are very proud of their work. The orthography represents their Chong
identity as members of a unique ethnic community of Thailand.

Figure 2: Thai alphabet Chart

Figure 3: Chong alphabet chart

Figure 4: Chong registers
Using one word for one symbol, the practical orthography using a Thai-based script is simple. Once
the tool for writing has been developed, the language speakers can start writing stories (at different difficulty
levels), editing (story writing, language), illustrating and book binding. Reading materials in the local
language have therefore been produced by various members of the community.
Since it is a dream of the Chong to have their language taught in the school system, the curriculum
for teaching Chong as a subject has been developed. In this way, Chong students are able to use Chong as
well as Thai (official language) in an educational atmosphere. They are proud of their part in Chong
Language Revitalization Program (CLRP). The preparation of teaching materials is in accord with the
cultural calendar and localized content.
For teaching Chong in school, teachers were selected from among the people in the community. The
teacher selection criteria are based on proper pronunciation of Chong, dedication to the CLRP, and
acceptance by the Chong community.
Apart from language classes in school, field trips are organized for the students to the community
forest to learn about plants, animals, beliefs, food items etc., that are important to the local culture. The
cultural activities are organized for community at large on cultural days, as well as for students to learn at the
Chong Community Learning Center. For example, they learn to cook typical Chong food, as well as Chong
dessert.
The initial success of CLRP, which is the first language revitalization program in Thailand, is quite
encouraging. It is strongly empowering for the community. This program has revitalized not only the Chong
peoples language, but also their self-confidence and self-esteem. As a way of promoting minority language
education in school, it has become a model for other struggling, endangered groups such as the Nyah Kur, So
(Thavung), Lavua etc. Additionally, the Chong students have been able to pass the National Standard Test in
Thai and Math for the first time. This project has also contributed to a reconsideration of education policy. It
is actually the first cooperative activity between linguists, education experts and the community to carry out a
revitalization program for an endangered language.
3.2 The last breath to revitalize Kasong, a language in the last stage of endangerment before extinction.
As for, the Kasong, despite very few good speakers, the people would like to relearn their language
with help from linguists; whereas Chung and Samre have no hope for survival after the death of the elderly
speakers. They have neither the energy nor the enthusiasm of the young to preserve their language. However,
there is a woman over 50 years of age who is still a fluent Kasong speaker, who is very much interested in
collecting Kasong language and local wisdom. She wants to teach the language to the younger generation and
become the key person for the Kasong revitalization project.

3.3 Pearic Languages, documentation and revitalization attempts
Peaic Languages and Stage of Endangerment
Language Revitalization
Activities CHONG
(7)
CHUNG
(Sa-oc)
(8)
KASONG
(8)
SAMRE
(8)
1. Community study / Language
Situation survey / language vitality
/ Attitude Survey
-- --
2. Orthography Development -- --
3. Literature/book in Vernacular
Language
-- -- --
4. Curriculum Development for
teaching Ethnic Language (in
School)
-- -- --
5. Learning-Teaching-Reading
Materials
- Primer
- TPR lessons
-- -- --
6. Community Learning Center
(outside school)
-- -- --
4. Conclusion
The entire Pearic branch of AA is dying. Even though there are record numbers of Pearic speakers in
various locations in Cambodia, they are not easily found these days. Kasong, Samre and Chung that are
found in Thailand are at the last stage of endangerment (according to Fishmans GIDS) have very little hope
of survival after the death of the last few elderly speakers. Chong which at the stage before last still have
more and enthusiastic speakers who want to preserve their language and have joined hands with Mahidol
linguists to reverse the situation. The lexical and syntactical characteristics of Pearic languages which are
heavily influenced by the dominant (Thai) language as well as the register complex and tonogenesis found in
Chong, Chung, Kasong, and Samre have been documented. While Chong and Chung are register languages,
Samre is a tone language, and Kasong is at transitional stage of developing tones. Even though the Chong
language revitalization process is considered to have started twenty years too late, the teaching of Chong as a
subject in the formal school system, as part of the Chong language revitalization, has been rather successful
and become a model for other languages with the same problem. On the other hand, for the Kasong, which is
at the last stage, their descendants are trying to relearn and document their ethnic language as much as
possible. Only a miracle can help preserve this language beyond the current generation even though they
have developed a writing system as a tool for learning and documenting their language. In general, the Pearic
speaking people have gradually given up their way of life and live as the dominant group does. The Chong
children grow up with limited exposure to Chong language at home but learn various aspects of Chong
language and culture by attending classes and other organized events. These languages are likely to become
extinct by the end of this century. What can we do from the larger society? Apart from language
revitalizing, studying and documenting the language before its lost forever, we can help with the
development a national language policy that supports the use of indigenous languages in public, in school,
and in mass media in their own area alongside the official / national language and international languages)
for the sustainability of the preservation activities.

References
Diffloth, Grard. (1984). Davaravati Old Mon Languages and Nyah Kur. Bangkok: Chulalongkorn
University Press.
Fishman, Joshua A. (1991). Reversing Language Shift. Clevedon, England: Multilingual Matters.
Hinton, L. and K. Hale. (eds.). 2001. The Green Book of Language Revitalization in Practice. San Diego, CA:
Academic Press.
Isara Choosri. (2002). Mapping dialects of Chong in Chanthaburi Province, Thailand: An application of
geographical information system (GIS). M.A. thesis, Institute of Language and Culture for Rural
Development, Mahidol University.
Isara Choosri. (2007). Investigating contact-induced language change: case of Chung (Saoch) in Thailand
and Cambodia. Ph.D. dissertation. Research Institute for Languages and Cultures of Asia, Mahidol
University.
Krauss, Michael. (1992). The worlds languages in crisis. Language 68(1): 4-10.
Malone,Dennis. (2003). Developing curriculum materials for endangered language education: Lessons from
the fiel. International Journal of Bilingual Education and Bilingualism 6(5):332-348.
Matisoff, J. A. (1991). Endangered languages of mainland Southeast Asia. In K.H.Robins and E.M.
Uhlenbeck (Ed.s). Endangered Languages, 189-228. Oxford University Press.
Nattamon Rojanakul. (2009). Chong syntax. M.A. thesis, Research Institute for Languages and Cultures of
Asia, Mahidol University.
Noppawan Thongkham. (2003). The phonology of Kasong at Khlong Saeng village. M.A. thesis, Institute of
Language and Culture for Rural Development, Mahidol University.
Pornsawan Ploykaew. (2001). Samre Grammar. Ph.D. dissertation. Institute of Language and Culture for
Rural Development, Mahidol University.
Siripen Ungsitipoonporn. (2001). A phonological comparison between Khlong Phlu Chong and Wang
Kraphrae Chong. M.A. thesis, Institute of Language and Culture for Rural Development, Mahidol
University.
Sirirat Srisombat. (2009). A community learning center and the revitalization of the Chong ethnic identity.
M.A. thesis, Institute of Language and Culture for Rural Development, Mahidol University. (in Thai)
Smally, William. (1994). Linguistic Diversity and National Unity: Language Ecology in Thailand. Chicago:
Chicago University Press.
Sunee Kamnuansin. 2002. Kasong syntax. M.A. thesis, Institute of Language and Culture for Rural
Suwaphat Chanmekha. (2003). The hidden language: a case study of Kasong language attitudes, uses and
abilities in three villages in Trat province. M.A. thesis, Institute of Language and Culture for Rural
Suwilai Premsrirat. (1996). Phonological characteristics of So (Thavung), a Vietic language of Thailand.
MKS 26:161-178.
Suwilai Premsrirat. 2000. Developing writing system for the Chong language. JLC 19.2:5-18 (in Thai).
Suwilai Premsrirat. (2001). The future of Nyah Kur. In Robert S. Bauer, ed. Collected Papers on Southeast
Asian and Pacific Linguistics. pp. 155-165. Canberra: The Australian National University.
Suwilai Premsrirat and Dennis Malone. (2005). Language development and language revitalization in Asia.
Mon-Khmer Studies. 35:101-120.
Suwilai Premsrirat. (2006). Thailand: Language Situation. Encyclopedia of Language and Linguistics. 2nd
Edition. Edited by Keith Brown. Oxford: Elsevier.
Suwilai Premsrirat. 2007. Endangered Languages of Thailand. International Journal of Sociology of
Language. 186(2007): 75-93.
Suwilai Premsrirat, Isara Choosri ans Ekapong Suwankases (2001). Ethnolinguistic mapping of Thailand.
Salaya: Mahidol University.


Appendix

Photograph 1-2:Pa Cin and Nai Chian, fluent speakers of Northern Chong dialect

Photograp 3-4: A Northern Chong speaking grandmother and a grandchild who cannot speak the language
(Left). Mr.Kasem, the last fluent speaker of Southern Chong dialect (Right).

Photograp 5-6: The Chong speakers working with professional linguists (Left). The last three speakers of the
Eastern Chong Dialect (Right)

Photograp 7-8: The well-known Kamnan Chern, the first Chong project leader (Left) The middle age key
persons of Chong revitalization program (Right).

Photograp 9-10: Two of the last few speakers of Samre and Kasong (Left). A Kasong- speaking grandmother
and her grandson who cannot speak the language (Right).

Photograp 11-12: Chung Speakers in Cambodia (Left). Chung Speakers in Thailand (Right).

Photograp 13-14: Thai based Chong language development

Photograp 15 17: Writing stories and producing teaching materials in Chong.

Photograph 18: An example of a Big Book in Chong.

Photograph 19: Curriculum development for teaching Chong as a subject in school.

Photograph 20 - 23: Learning to read and write in Chong.

Photograph 24 - 25: Learning more Chong vocabulary during the trip to the forest.

Photograph 26: Thai based Kasong writing system.

Photograph 27 - 28: Developing Kasong writing systems using Thai alphabet.

Photograph 29 - 30: Pa Somsri, the last Kasong active speaker.

Photograph 31 - 34: Elderly Kasong speaker and younger Kasong descendants in their try to transfer
knowledge.

Pucilowski, Anna. 2011. Aspects of Ho Phonetics and Phonology. In Sophana Srichampa and Paul Sidwell (eds.)
154
Aspects of Ho Phonetics and Phonology
Anna Pucilowski
*

University of Oregon
1 Introduction
The phonetics and phonology of Ho, a North Munda language of central-eastern India, have not
been well described to date. This paper is a more comprehensive description of that system than previous
works (e.g., Deeney 2002, Anderson et al. 2008). We will see that some of the more interesting features of
Ho include the phonetic prenasalization of word initial stops, as well as vowel harmony. In the rest of this
section I briefly introduce Ho. Section 2 covers the consonants and section 3 describes the vowels. In section
4, I discuss syllable structure and section 5 looks at suprasegmental features, including vowel harmony.
Ho is spoken in the East Singhbum district of Jharkhand and the Mayurbhanj and Keonjhar districts
of the state of Orissa, India. There are approximately 1,500,000 Ho speakers (Lewis 2009). Ho is very closely
related to Mundari, to the extent that some researchers have called Ho and Mundari dialects of the same
language rather than separate languages (Pinnow 1959, cited in Osada 2008). According to Anderson et al.,
(2008) there is about 80-85% similarity between the two languages, at least for the Mayabhanj dialect (Osada
2008:161).
There are two known dialects of Ho. The Mayurbhanj dialect of Orissa has been much less studied
than the Chaibasa dialect of Jharkhand. According to Anderson et al., there is some variation between the
dialects in the vowel harmony (2008:199) and in the pronunciation of certain consonants (2008:201). More
differences may come to light as the Mayurbhanj dialect is studied further.
1.1 Speakers for this study
The data for this sketch come from Ho speakers in Jharkhand, who are mostly from the Chaibasa
area but living in Ranchi, Jharkhands capital. The samples examined for this paper come primarily from one
speaker. He is a university educated 30 year old who speaks Ho, Hindi, Mundari, Santali, Bengali, Oriya and
English. The elicitation sessions were conducted primarily in English, translating into Hindi as necessary.
1.2 Methods
We collected approximately 2500 words over a three month period. The words were collected for a
Talking Dictionary Project. Words were recorded once with their English translation as .wav files onto an
Olympus LS-10 sound recorder at a sampling rate of 44,100Hz using the recorders internal microphone.

*
The data for this paper were collected during fieldwork made possible by Living Tongues Institute for
Endangered Languages. Special thanks also to Ganesh Murmu, Bhubneshwar Sawaiyan and the Ho students from the
Department of Tribal and Regional Languages at Ranchi University.
Aspects of Ho Phonetics and Phonology 155

2 Consonants
The phonemic consonant inventory of Ho is shown in table 1
105
. There are either 21 or 22 phonemic
consonants, depending on the analysis of the palatal nasal (see below). The phoneme /n/ has allophones [n],
[] and []. In addition, the palatal and velar nasals, included in angle brackets, represent just one underlying
phonemic consonant. This is discussed further in section 2.3.

Labial Dental Retroflex Palatal Velar Glottal
Stops p, b t, d , c, k, g
Fricatives s h
Nasals m n < >
Flaps r
Lateral l
Glides w j
Table 1: Consonant Inventory of Ho
In the rest of this section I will discuss the consonants according to manner of articulation.
2.1 Stops
As seen in table 1, Ho has five stop points of articulation, including the retroflex stops, with voiced
and voiceless counterparts. There is also a glottal stop.
Table 2 shows average VOT for the voiced and voiceless variants of three stops in Ho. I restricted
measurements to an #_a environment which meant that there were not enough tokens of // and // to measure
(retroflex stops only appear in borrowed words). Additionally, I omitted tokens of /c/ and // because they are
phonetically too fricative-like to measure VOT.
stop (no. of tokens) average VOT time (ms) range
k (10) 49 38 64
p (6) 30 17 43
t (8) 25 14 23
g (10) -68 -46 -48
b (10) -89 -54 -119
d (10) -78 -47 -112
Table 2: Average VOT for Ho stops
We can see from the table that VOT for unaspirated voiceless stops is not long, with the velar stop
having the longest average duration at 49ms. We may treat Ho voiceless stops as short-lag stops (stops with
VOT of 0-35ms), with the VOT of /k/ somewhat longer. The short-lag voiceless stops may be one of the
factors in the prenasalization of voiced stops word initially (see in section 2.1.1).
Word initially and medially, there is always a phonemic contrast between voiced and voiceless stops.
The following examples of minimal pairs or near minimal pairs illustrate contrast between voiced and
voiceless stops.
p : b
/pura/
much, many
/bura/
to draw or ladle out
/capa/ draw picture /caba/ finish

105
I use IPA transcriptions in table 1 and throughout the paper
156 Anna Pucilowski

t : d
/tul/ carry, in arms /dul/ pour
/ata/ roast /ada/ know

: : : :
/u:/
squirrel
/ur/
quail
/kua/
straw
/kua/
roseapple

c :
/ci/ or
/i/
soul
/raca/ courtyard
/rea/
coolie

k : g
/kuni/
terraced, upland field
/guni/
cow
/hake/ axe /haga/ brother

We can also see contrast between the palatal and velar stops, as the following words show:

c : k
/cimin/ how much, many /kimin/ the wife of ones son
/-ici/ caus suffix /ikir/ deep

: g
/om/
eat /gom/ wheat
/saaw/
decorate
/saga/
type of grass seed

Thus far we have seen that the voiced and voiceless pairs of stops are contrastive in initial and
intervocalic position for each place of articulation. There are, however, a couple of instances of free variation
in point of articulation. According to Anderson et al. (2008), there is individual variation among Mayurbhanj
Ho speakers between [d] and [], and [t] and [c], so that /cimi/ how many may also be pronounced [timi].
Similarly, how do you say may be said as [tilekepe kadije] or [cilekepe kaije] (Anderson et al. 2008:201).
The only examples they give of these consonants occur before [i] so it may be that /i/ is causing the
palatalization in some words and therefore that /t/ and /d/ might be the underlying consonants.
The following examples demonstrate initial and medial contrast between the dental and retroflex
stops, both voiced and voiceless.
t :
/mata/ ripe
/maa/
whey
/tu:/
mulberry
/u:/
squirrel

d :
/duku/ sorrow
/uki/
urine
/didi/ a vulture
/ii/
to stand on tiptoes

There is however, free variation between /t/ and // and /d/ and // in some words, as in the following
examples.

(1) [tai] [ai] stay, remain
[data] [ata] tooth
[dani] [ani] a small handle, connecting device
There are minimal pairs establishing the phonemic status of // and //.
: : : :
/bua/
tree trunk
/bua/
old man
/ui/
defile
/ui/
friend, companion
/goa/
whole
/goa/
ground, land

Again, Anderson et al. (2008:202) note that it is also possible to get some free variation between //
and //, and give the following examples from Deeney (2005).
(2) [pee] [pee] pluck twig or small branch with one or both hands
[ka:ob] [ka:ob] crab
[poa] [poa] intestines (Deeney 2005)
The fact that // can vary freely with both /t/ and // in some instances leads Anderson et al. (2008) to
question the phonemic status of //. Given that there are minimal pairs, we must posit it as a phoneme at this
stage. It seems that there are more instances of free variation intervocalically than initially. In fact, there is
only one instance of variation initially ([tai] ~ [ai] stay, live). Retroflex sounds do however, only occur in
borrowed words only so it may be the case that they can vary with non-retroflex sounds.
These instances of free variation are interesting and need further research. These may be instances of
regional variation. In the case of dental/palatal variation before /i/, it may be a case of palatalization, and so
not really free.
2.1.1 Prenasalization of voiced stops/phonetic variation in stops
There is a tendency for the voiced stops {b, d, g, , } to be prenasalized, e.g., /bandor/ [
m
bandor]
monkey in word initial position. The nasal that appears before the stop is homorganic and not normally
equal to other nasals in intensity or length. Figure 1 shows a spectrogram of a prenasalized voiced stop in
[
m
banda] pond, tank, where we can observe the nasal band that precedes the stop. We can see that the
prenasalized stop is comparable in intensity to the second nasal portion ([n]) of this word. Not all pre-nasal
portions were as intense as in this instance.

Figure 1: Spectrogram of [
m
banda] tank, pond
158 Anna Pucilowski

Prenasalization does not happen in every instance of a voiced stop. However, it can be observed in
all of my speakers to some extent. Table 3 shows the proportion of words containing an initial voiced stop
that were prenasalized.
b d g

prenasalized 91% (247) 78% (134) 90% (165) 58% (107) 77% (20)
not prenasalized 9% (23) 22% (38) 10% (19) 42% (78) 23% (6)
total no. of tokens 279 172 184 185 26
Table 3: Percentage of prenasalized initial voiced stops in Ho words
I coded all of the word initial voiced stops in my data as either prenasalized or not prenasalized
based on auditory impression. The tokens come from five speakers, three males and two females, all under
the age of 30. We can see from the table that voiced stops in initial position are prenasalized more often than
not. The instances of // being prenasalized are much fewer. This is most likely due to its more fricative-like
nature. Cross-lingistically fricatives can be pre-nasalized, however, it is less common.
Two of the speakers (one male, one female) had fewer instances of prenasalization than the other
three. However, they still produced prenasalized stops in some instances. It is possible that the
prenasalization of initial voiced stops varies depending on social factors such as region and education. Given
that all my speakers were under 30, future research must address whether the prenasalization is more
common among younger speakers and therefore an indication of language change or whether all Ho speakers
produce prenasalized stops and it is a stable phonetic feature.
The prenasalization of voiced stops is not always an indication of phonological change in progress.
From an articulatory point of view, a prenasalized stop is just one maneuver which speakers can use to
facilitate voicing during the stop closure and it serves to highlight the voicing in the stop (Johnson 2005:139,
see also Ohala 1997).
It is difficult to see from the data whether the voiced stops are pre-nasalized when they occur
medially. Ohala predicts that inter-vocalic voiced stops may spirantize (1997:95) and we do have some
examples of that with some speakers in Ho, at least with the bilabial stop.
(3) [jibon] [jion] life
[babata] [baata] scabies, itch
Whether intervocalic voiced stops spirantize regularly in Ho needs to be further investigated.
2.1.2 Word final stops
We now turn to stops in word final position. Voiceless stops do not appear in word final position,
except in some borrowed words, e.g., /kek/ cake, /sut/ suit, salwar kameez and /biskut/ biscuit, /kop/
cup (from English) and /camac/ spoon (from Hindi). I have no examples of word final //.
Voiced stops /b/ and // appear phonemically word finally but there appears to be some phonetic
variation in how they are pronounced. They are frequently unreleased or they are preglottalized and
accompanied by a nasal release.
(4) /la/ [la] or [la

n
] bread
/porob/ [porob] or [poro
b
m
] feast, festival
In Ho, both monosyllabic and polysyllabic words can have an audible nasal release in citation form.
However, in connected speech, the nasal release can only be detected when it occurs at the end of an
intonational phrase, elsewhere the stop is merely unreleased.
The voiced stops /d/, /g/ and // do not appear word finally except in borrowed words, e.g., /ne:g/ a
ritual, rite (from Hindi). However, there is also a glottal stop phoneme that appears word finally. Historically,

this is thought to be an allophone of /g/ (Anderson et al. 2008:200) but this is not clear synchronically.
106
The
glottal stop is always followed by an echo vowel which has the same quality as the preceding vowel.
(5) /da/ [daa] water
/seta/ [setaa] morning
I represent these words phonemically with the glottal stop, e.g., /da/ dog, not /dag/. Words with a
final glottal stop are contrastive with words that end in a simple vowel, either long or short, as in the
following examples.
(6) /iju/ shout, call out /iju/ fall from a height
/o:/ fruit /o/ sweep
As in the examples with the nasal release, the echo vowel can be heard in the citation form of
polysyllabic words, such as /seta/ morning as well as in monosyllabic words. Given that the echo vowel is
entirely predictable, it is not normally written in phonemic transcriptions.
2.2 Fricatives
There are only two fricatives in Ho: /s/ and /h/. /s/ can appear in onset and coda position, although
most of the instances of syllable final /s/ seem to be in borrowed words (/inis/ thing being an exception).
(7) /sarkam/ leaf
/rasi/ juice
/h/ can only appear in onset position.
(8) /hisi/ twenty
/hende/ black
In some words, it seems that [s] is in free variation with [], e.g., [sani] or [ani] rooster, [hoso]
or [hoo] goose. This could be particular to certain speakers however. The bilabial fricative allophone
was mentioned above.
2.3 Nasals
Five nasal consonant sounds are attested in Ho, but they do not all have phonemic status. The
bilabial and dental nasals are the only nasals that can appear in all positions, as shown in table 4.
m n
/mana/ forbid /nama/ new
/hambal/ heavy /hende/ black
/nimin/ contents /guni/ wise
/ikum/ kneel /isin/ cook
Table 4: /m/ and /n/ in Ho
There are three other nasals that probably represent two phonemic consonants. First, the retroflex
nasal [] has a very restricted distribution and is best analyzed simply as an allophone of /n/. It only appears
intervocalically in very few instances, mainly loan words, e.g., [dua] resin of a sal tree used to make
incense (Anderson et al. 2008:202). Otherwise, [] only appears before // and // so might therefore be
treated as an allophone of /n/.
(9) ['mai] food
['gai] small bell

106
In Mundari, Osada also regards the glottal stop as an allophone of both /g/ and /j/ (Osada 2008:102).
According to him, /g/ appears after /a/ and /j/ after the other vowels.
160 Anna Pucilowski

Secondly, the velar and palatal nasals seem to be in complementary distribution, suggesting that they
represent just one phoneme underlyingly. Neither appears in onset position or intervocalically. I do not,
however, analyze all instances of [] as deriving from the same phoneme. First, we see examples of the velar
nasal appearing before the velar consonants /k/ and /g/.
(10) ['sigi] sun
['taku] pit (of fruit)
Given that this is entirely predictable, we can treat this [] as an allophone of /n/. These instances of
[] are normally represented as n in both Devanagari and romanized writing, which is further evidence of
[]s allophonic status here.
However, there is also an [] which appears word finally and thus seems to contrast with /m/ and /n/
while it is in complementary distribution with the palatal nasal [], as we see in the following table.

n

[isin] cook
[tisi]
today
[heben] bitter
[sete]
spring, fountain
[oan]
bathe, wash
oneself
[a]
I [era] scold
[ibon]
life [holo] flour
[susun] dance [sasa] tumeric
Table 5: Alveolar, palatal and velar nasals in word final position
As we might expect, the palatal nasal appears mostly after the front vowels {i, e} and the velar nasal
normally appears after the back vowels {u, o, a}:

[bete]
wait [unu] play
[sete]
spring, fountain [ro] color
[bi]
snake [sasa] yellow
[tisi]
today [gono] orphan
Table 6: Palatal and velar nasals in Ho
However, there are some exceptions to this tendency. Anderson et al. (2008) claim that both [] and
[] can follow [u] (2008:202). Indeed, the only minimal pairs appear to be following [u].
(11) [apu] father - vocative form [apu] my father
[ru] to husk [ru] sensation of having a
limb asleep
They claim that this is evidence of phonologically front vs. back [u] (Anderson et al. 2008:202). I
measured the formants of [u] for [ru] to husk and [ru] sensation of having a limb asleep as spoken by
my consultant and there was no significant difference between them. The F2 of the [u] in [ru] was 1101Hz
and in [ru] it was 1191Hz. The palatal nasal had a slightly higher F2, and therefore fronter vowel but we
need more tokens of [u] plus the palatal and velar nasals before we can draw any conclusions about a front vs.
back [u].
In addition, there is one minimal pair involving [a]:
(12) [a] dawn [a] I

Anderson et al. record [germo] a fleeting smile as another exception, because it has a palatal nasal
following the back vowel [o]. They also note a tendency for speakers of Mayurbhanj Ho to have [i]
generally rather than the [i] of Chaibasa Ho, e.g., [tisi] today rather than [tisi] (2008:202).
An additional question is how the nasal consonants interact with vowel harmony (see section 5.1). A
low vowel [a] frequently raises to [e], when it appears after a high vowel. We might then ask whether a
following [] changes to [] after a higher fronter vowel. In the only example I have where that rule might
apply, the nasal appears to stay the same. /dija/ rice beer becomes [dije] phonetically, and the nasal does
not seem to change. In another example, [imina] enough, we expect the final [a] to raise to [e]. It would be
interesting to see if the nasal consonant in turn changes. Unfortunately, in the examples we have, it is always
followed by =ge, an emphatic clitic. The velar stop [g] seems to prevent the nasal from changing.
I here make some tentative suggestions about the number of nasal phonemes in Ho. We can posit
three nasal phonemes: /n/, /m/ and maybe //. /n/ has three allophones: [n], [], which appears before
retroflex stops and [], which appears before velar stops. I also posit // as a phoneme with two allophones,
[] and [] that appear word finally. [] occurs after the back vowels {u, o, a} and [] after the front vowels
{i, e} The phoneme could equally be //.
2.4 Liquids
There are three liquid consonants in Ho: /l, r, /. The /l/ is clear in all positions and the /r/ phoneme is
a flap. Both /l/ and /r/ can appear in all positions and they are contrastive.
(13) /kamal/ lotus /kamar/ blacksmith
/lo:/ burn /ro:/ dry
/alom/ net /arom/ egg
The retroflex flap // primarily appears intervocalically and very rarely finally. It contrasts
phonemically with /r/.
(14) /gai/ cart /gari/ rake together small shed
/haa/ bullock, steer /hara/ grow
/gaa/ river /gara/ cement, mud paste
/suga/ handsome
Although // is contrastive with /r/, it is sometimes in free variation with //, as noted above in
section 2.1. It is not clear whether this variation is intra-speaker or attributable to a particular regional dialect.
2.5 Glides
There are just two glides in Ho; /w, j/. Neither occurs word initially (although note that the Ho script
is sometimes called Warang Chiti) and only very rarely in final position, mostly in loan words. The labial
glide /w/ has a more restricted distribution than /j/. It only seems to occur after the back vowels {u, o} and
before /a/, as in (2.5):
(15) /towa/ milk
/guwa/ betel nut
The palatal glide /j/ seems to appear between all combinations of vowels, although I only have one
example of it appearing between a back vowel and /a/ (where we see /w/), /hjar/ father-in-law.
(16) /tuju/ jackal
/tajom/ after
/hojo/ air
/dija/ rice beer
162 Anna Pucilowski

Both glide consonants are inserted to ease pronunciation in certain contexts, as we will see in section
2.6.3 on epenthesis.
2.6 Other Phonetic Processes
2.6.1 Free variation
/l/ are /n/ are contrastive phonemes in Ho, as we can see in the following examples:
(17) /neka/ like this /leka/ count
/panti/ row /pali/ overturn
However, we can also see some examples of free variation between /l/ and /n/, as in (2.6.1).
(18) [nel] [lel] see
[nili] [lili] honey bee (also [lele])
If this variation is always between [n-l] and [l-l], it could be the result of long-distance assimilation
(or dissimilation). It might also be regional, rather than within-speaker. My consultants suggested that
speakers who use the [l]-variant are from Chaibasa town.
Another type of variation that is closer to true free variation is between // and /j/. This type of
variation can be heard within the same speaker and was also noted by Anderson et al. for Mayurbhanj Ho
(Anderson et al. 2008:203. These examples of j/ variation in /arom/ egg are spoken within phrases of each
other, by the same person.
(19) /..mendo agu-ke--a bi arom.
...but bring-T/A-TR-FIN snake egg.
Endo bi jarom-a agu-le--redo.../
So snake egg-POSS bring-T/A-TR-WHEN/IF
...but he brought a snake egg. So when he brought the snake egg...
(Girl-snake, lines 17-18)
2.6.2 Metathesis
We saw in the previous section that there is sometimes free variation with [n] and [l] in Ho. As well
as this free variation, we also see metathesis with these consonants, as in example (20), where either [n] or [l]
switches places with [r].
(20) [rulbi] [nurbi] milk-snake (Anderson 2008:204)
In the following example of metathesis (21) we see consonant variation between [k] and [r] and also
that the order of the consonants switches, so that [s] precedes [k] in [maskal], but follows [r] in [marsal].
(21) [maskal] [marsal] shine (of ame) (Anderson 2008:205)
Other lexical variants where metathesis has applied include:
(22) [lapa-lapa] [lagab-lagab] out of breath
[siimi] [simii] bean
In the first example above, we notice that the bilabial stop is voiceless intervocalically, but voiced
word finally. Also note that when the velar nasal appears word medially, it must be syllable final rather than
syllable inital and so requires a homorganic stop to begin the next syllable.
Anderson et al. also note that metathesis can sometimes be obscured by infixation. For example, the
reciprocal infix -p- can produce [bepea] ~ [tepeba] meet from /bea/ reach (Anderson 2008:204).

2.6.3 Epenthesis
There are various words in which either /j/ or /w/ is inserted in order to ease pronunciation. This is
most frequently between morphemes, as in the following examples.
(23) [maasor] master (Anderson et al. 2008:204)
(24) /ako/ 3.pl + /a/ poss [akowa]
In general, /j/ follows a front vowel and /w/ follows back vowels. After /a/, neither glide is necessary.
According to Deeney, the glides that are inserted as epenthesis are normally written by Hos when writing
Devanagari and he also uses them in his romanized script (Deeney 2002:xix).
3 Vowels
Ho has five contrastive vowel positions, but with two additional contrastive features: length and
nasalization. Vowels can be short or long and nasalized or oral. Vowels can also be glottalized, as described
in section 2.1 as a feature of certain word final glottal stops. Hos vowel harmony system will be described in
section 5.1.
3.1 Short Vowels
As mentioned, Ho has a five vowel system; /i, e, a, o u/. Table 7 shows example words to
demonstrate that all five oral vowels are contrastive in Ho.
/miai/
sweetmeats
/cii/
a letter /bita/ length between the
tip of the thumb and
the tip of one finger
/meta/ to say to /cetan/ above
/bea/
to arrive, reach
/mata/ to ripen
/caa/
to split, crack
open
/bage/ leave, abandon
/moa/
thick, fat /coka/ frog /boka/ a stupid person
/mua/
nose
/cuu/
mouse
/bua/
tree trunk
Table 7: Contrastive short vowels in Ho
The vowel plot in figure 2 shows measurements of F1 and F2 for 10 tokens of each vowel. The
vowels all occurred in the first syllable of words of the type CV.CV (stress normally falls on the first syllable
of this type of word, see section 5.2). They all appear between obstruents, mostly stops, but also some
fricatives. Formant measurements were taken at the mid-point of the vowel. All tokens were recorded by the
same male speaker.
164 Anna Pucilowski

Figure 2: Short vowels in Ho
We can see from the vowel plot that the vowels in Ho have a similar distribution to what we might
expect from a language with a 5-vowel system. Table 8 shows the mean formant values for the short vowels
from figure 2 in Ho.
F1 (Hz) F2 (Hz)
i 311.3 2042.92
e 426.79 1856.29
a 592.29 1286.73
o 455.2 1021.58
u 325.33 1034.22
Table 8: Mean formant values for F1 and F2 in short vowels in Ho
There are two salient instances of allophonic variation of vowels in Ho, as well as some instances of
vowel neutralization. The first is [e] as a raised variant of /a/, as shown in (25). The [e] allophone occurs only
word finally in unstressed position after the high vowels {i, u}, as a type of harmony. Anderson et al. notice
this for Mayurbhanj Ho and it is also a feature of Chaibasa Ho.
(25) /kula/ ['kule] tiger
/luga/ ['luge] nest
/misa/ ['mise] once
but,
/gaa/ ['gaa] river
/boa/ ['boa] load, bundle
/seta/ ['seta] dog
This is also discussed in section 5.1 on vowel harmony.
The second type of allophony in vowels is the nasalization of vowels when they precede a nasal
consonant, as in the following examples.

(26) /enga mini/ [ga mi] ewe
(27) /hila/ [hil] disgust
As we will see in section 3.3, nasalization is also a phonemic feature of Ho vowels.
There are also examples of vowel neutralization. A schwa vowel [] can occur in unstressed position,
mostly word internally.
(28) /kakala/ ['kakla] to shout
/tisi/ [t'si] today
/dudulum/ [dud'lum] pigeon
This is not particular to any vowel and it raises questions about stress patterns in Ho. It seems that
Ho has three types of vowels: stressed, unstressed and reduced, and unstressed and unreduced (as in the first
syllable of [dudlum] pigeon). This leads us to ask whether Ho is a stress-timed language. At this stage it is
not clear but future research will look at stress patterns and their relationship to vowel reduction, while also
paying special attention to the interaction of stress with Hos complex morphology.
3.2 Long Vowels
All five Ho vowels can be either short or long. Figure 3 shows a vowel plot for 10 tokens of each
long vowel. There are many fewer long vowels in my data than short vowels, so these tokens come from a
variety of phonetic environments, including both open and closed syllables. There are also some tokens from
another speaker, although still male.

Figure 3: Long Vowels in Ho
There is some debate about whether vowel length is phonemic in Ho. Deeney (2002:xiii) claims that
it is indeed phonemic. And, as we see from table 9, there are several minimal pairs that suggest this is true in
the Chaibasa dialect.
[kani] a pointed edge [ka:ni] story
[me]
eye
[me:]
iron
[agu] bring [a:gu] lower
[cera] diarrhea [ce:ra] beautiful
[gom] wheat [go:m] to accompany someone
Table 9: Long vs. short vowel minimal pairs in Ho
166 Anna Pucilowski

Zide (1991:537) and Anderson et al. (2008) state that vowel length is not phonemic in Ho. Zide
claims that vowel length is in fact geminate. There is no synchronic morpheme boundary in the long vowels
so there does not seem any advantage to positing that they are geminate rather than long.
Table 10 below shows the average length of long vowels versus short vowels in Ho, measured in
milliseconds. I measured 10 tokens each of both long and short vowels. As stated above, there were fewer
long vowels, thus the long vowels are from both open and closed syllables of varying word lengths, while the
short vowels are all in the first syllable of words with CV.CV shape. We can see from the table that, on
average, long vowels are more than twice as long as short ones in Ho.
Short Long
range mean range mean
i 51-101 ms 71 ms 102-316 ms 203 ms
e 55-102 ms 71 ms 152-323 ms 212 ms
a 54-106 ms 80 ms 140-280 ms 202 ms
o 64-98 ms 83 ms 145-274 ms 217 ms
u 44-91 ms 75 ms 131-352 ms 203 ms
Table 10: Comparison of Length of Short and Long Vowels in Ho
Historically, at least some of the long vowels in Ho seem to result from the loss of an []. If we
compare Mundari and Ho vocabulary (table 11), we can see that in some words where Mundari has an //, Ho
has lost the consonant and is left with a long vowel.
Mundari Ho Gloss
[hoo]
[ho:] man
[duum]
[du:m] sleep
[sepeed]
[sepe:d] young man
[re%e%] [re%:]
joy in the company of others
Table 11: Comparison of Ho and Mundari vocabulary, from Deeney (2002:132-133)
Given that long vowels are distinguished in production, and the fact that we have minimal pairs for
every vowel, we must posit that vowel length is now a phonemic feature of Ho. The relatively low frequency
of long vowels cannot bear on their phonemic status.
3.3 Vowel Nasalization
In section 3.1, we saw that vowels are normally nasalized when they precede a nasal consonant.
Nasalization is also a phonemic feature of vowels in Ho. Some examples of nasalized vowels contrasting
with oral vowels are shown in table 12.
/ija/ grandmother
/ja:/
humid, moisture
/-rejo/ even if, although (verbal suffix) /re%jo/ creak, squeak
/bal/ to burn a hole into /bl/ flood
Table 12: Nasal and oral vowels in Ho
Long vowels can also be nasalized, however there are fewer examples of these. Some are shown in
(29).
(29) /s
:/ foul smell, stench

/:/ to erase, extinguish
/r:sa/ joy, delight
/d:si/ thirty
/:r/ leather, hide

To summarize thus far, vowels in Ho can contrast in both length and nasalization so that both long
and short vowels can be nasal or oral. Glottalization is also considered a feature of Ho vowels. This was
discussed in section 2.1 as it is a result of word final stops.
3.4 Diphthongs
There are some instances of two vowel sequences in Ho that might be called diphthongs.
(30) /bai/ work, build
/hau/ red ant
/umbui/ glutton
The vowels in these examples maintain a smooth transition between the targets, and are longer than
short vowels. I measured the lengths of the diphthongs and the results are presented in table 13.
Diphthong Mean length (ms) Range No. of tokens
/ai/ 223ms 126-292 11
/au/ 216ms 150-330 6
/ui/ 206ms 154-305 11
Table 13: Mean length of three diphthongs in Ho
We can see from the table that the average length of these two vowel sequences is about the same as
the long vowels in Ho (see table 10), although, like the long vowels, there is a lot of variation. The variation
seems to be dependent on whether the vowels occur in a closed or open syllable.
According to Deeney, Ho does not have diphthongs because two juxtaposed vowels normally retain
their independent sounds. He argues that these must be treated as independent vowel sounds because each
can be lengthened independently of the other, e.g., /bai/ make plus the inanimate marker, -i, gives us [bai:],
while adding the habitual marker lengthens the first vowel and gives us [ba:i] (Deeney 2002: xvii).
The vowel sequences above in (30) where there is a smooth transition between the vowels contrast
with instances where both vowels retain their individual sounds, and rather than a smooth transition, we see a
hiatus and sometimes a glottal stop between them. The following words demonstrate this:
(31) /toroe/ ashes [toroe]
/moroe/ acid, sour [moroe]
/ae/ 3.sg [ae
e
]
Whether we call the two vowel sequences in (30) diphthongs or two-vowel sequences must depend
on our definition of a diphthong. However, we must note that they are phonetically different from the
examples in (31) which have a definite hiatus.
4 Syllable Structure
In this section we will see the basic syllabic structure of Ho words, as well as the phonotactic
restrictions on those syllables.
168 Anna Pucilowski

4.1 Basic Patterns
The patterns for Ho syllables in words are shown in table 14.
Monosyllabic
VC /r/ hide, leather, animal skin
VVC
/a:/
herb
CV /nu/ drink
CVV
/u:/
squirrel
/bai/ work
CVC /cur/ surround, gush
CVVC
/u:r/
smooth
Disyllabic
V.CV /u.ku/ hide
V.CVC /a.bu/ wash hands and feet
CV.V /go.e/ wilt, die
CV.CV /ca.pi/ wash
CV.CVC /de.ra/ maybe
VC.CV /en.ka/ like that
VC.CVC
/a.kar/
sense
CVC.CV
/kun.u/
wooden post
CVC.CVC /ban.dor/ monkey
Trisyllabic
V.CV.CV /a.do.wa/ husked without boiling
V.CV.CVC /e.pa.ra/ quarrel
CV.CV.CV /sa.sa.ti/ torment
CV.CV.V /ku.la.e/ rabbit, hare
CV.CV.CVC
/ko.lo.wa/
to give a loan
CV.CVC.CV /go.pon.de/ quarrel, recip.
CV.CVC.CVC
/ka.ram.ca/
fox
VC.CV.CV
/an.gu.i/
finger
CVC.CV.CV /sin.du.ri/ vermillion
Table 14: Syllable Structure of Ho words
VV represents both long vowels and diphthongs in this table. I have only included examples of
monosyllabic words with VV sequences. However, long vowels and diphthongs also appear in all positions
in multi-syllabic words.
In Ho, both CVV and CVC syllables are treated as heavy, while only CV and V syllables are light.
This is evidenced by stress assignment (see discussion in section 5.2). Ho also has superheavy syllables: long
vowels and diphthongs can appear in closed as well as open syllables.
(32) ['a:ndi] marry, marriage
[ca:s] cultivate, farm
[kon.ai] put branches into an earthen pot to support something being steamed
['cau.li] uncooked rice

Table 15 is a summary of the phonotactic restrictions on the consonants in Ho.
Onset Coda Onset Coda
p
3
only borrowings s
3
only borrowings
b
3 3
h
3 7
t
3
only borrowings m
3 3
d
3
only borrowings n
3 3

3 3

7 3

3 3

7 3
c
3
only borrowings

7 3

3
only borrowings r
3 3
k
3
only borrowings

7 3
g
3
only borrowings l
3 3

7 3
w 3 (only V-V) only borrowings
j 3 (only V-V) only borrowings
Table 15: Phonotactic restrictions on Ho words
Note that table 15 only contains information about whether a consonant can occur in onset or coda
position or not. The phonemic status of various sounds is discussed in the relevant sections.
The glide consonants, /j/ and /w/, can appear in syllable initial position, however they do not occur
in word initial position, e.g., /tuju/ jackal, /guwa/ betel nut. The only other sonorants that can appear in
word and syllable initial position are the nasals /m/ and /m/, as well as /r/ and /l/.
We should also note that sonorants seem to be preferred in coda position. There are some exceptions
to this, and we see a few words with either //, // or /b/ in syllable final position, as in (4.1).
(33) /eka/ wicked, bad
/uka/ witchcraft
5 Suprasegmental Features
5.1 Vowel Harmony
Ho has vowel harmony based on height. The high vowels /i/ and /u/ do not occur with the mid
vowels /e/ and /o/ in a single word, with some exceptions. According to Deeney (2002:xviii), it is always the
case that the mid vowels raise to harmonize with high vowels rather than vice versa.
Borrowed words from other languages, such as Hindi and English, are usually adjusted to conform
to the vowel harmony rules.
(34) pothi book (Hindi) becomes [puti] in Ho
(35) police becomes [pulis] in Ho
There are only a few rare exceptions where high vowels and mid vowels are combined within a
single word. One very common one is [esu] very (although note that it has a variant [isu] that does follow
the harmonic pattern).
The second type of exception is due to the harmony that occurs when /a/ follows a high vowel and
raises to [e]. The only low vowel in Ho, /a/, might be called a neutral vowel because it can occur with any of
170 Anna Pucilowski

the other four vowels in a single word. However, /a/ does seem to raise or centralize to [e] or [] in some
words with /i/ or /u/ (see also section 3.1).
Figure 4 is a vowel plot showing the formant values for the raised /a/ vowel, alongside the other
vowel measurements from section 3. I measured twelve tokens of /a/. All occurred finally after a high vowel
in a word with CV.CV-type structure. All the tokens were spoken by the same male speaker.

Figure 4: Raised /a/ vowel in Ho
We can see from the plot that in most cases the raised /a/ has a more [e]-quality, and indeed overlaps
with many of the /e/ tokens. Although there are some instances where it is more like [].
Vowel harmony extends across certain morpheme boundaries in Ho so that some affixes will
harmonize with the vowel in the stem. We see that vowel harmony in Ho is normally progressive, as in
examples (36) and (37) but there are some examples of regressive harmony, as in example (35) above and
(38) below.
(36) /hasu/ sick +/o/passive [hasu:]
(37) /abu/ 1.PL.INCL +/leka/ like, as [abu-like]
(38) /en/ that +/ki/ dual [inki]
The aspect marker /-ke/ follows the verb root(s) in the clause and conveys something like perfective
aspect. When /-ke/ follows a root with a high vowel, it usually becomes [-ki] as we see in examples (40) and
(41).
(39) [om-ke--a] eat-T/A-TR-FIN
[men-ke--a] say-T/A-TR-FIN
[cike-ke--a] do-T/A-TR-FIN (/cika/ do)
(40) [tai-ki-n-a] stay-T/A-ITR-FIN
(41) [atom-idi-ki--redo] remove-take-T/A-TR-WHEN/IF (snake bite, line 23)

Further evidence that /a/ is a neutral vowel is shown in the third example in (39). Here we see that
the final /a/ of /cika/ raises to [e] and moreover, it serves to prevent the harmony from spreading from the
stem vowel to the affixes. We can call this /a/ an opaque neutral vowel (van der Hulst & van de Weijer 1995).
Vowel harmony in Ho appears to be sensitive to morphological structure so that some affixes are
more likely to undergo harmony than others. As we see in example (41) above, vowel harmony does not
always spread across the whole word: atom does not become [atum] and -redo does not become [rido]. In
terms of vowel harmony, atom-idi remove-take is a compound word, and therefore less likely to undergo
harmony (van der Hulst & van de Weijer 1995:501). The subordinating morpheme -redo is not part of a
compound, but it may be considered less close than tense/aspect markers such as /-ke/. Van der Hulst and
van de Weijer predict that derivational affixes are less likely to undergo harmony than inflectional affixes and
it does indeed seem that in Ho, harmony is sensitive to morphological structure.
Other affixes that appear to undergo harmony include /-tan/, a progressive marker (42), /-(e)jan/, an
intransitive past tense marker (43) and /-le/ a tense/aspect marker (44).
(42) [nam-tan-a] get-PROG-FIN, but [lel-ki-ten-a] look-DL-PROG-FIN
(creation myth, lines 23, 57)
(43) [hara-ejan-a=ki] grow-PST.ITR-FIN=dl, but [hapanum-ijen-a=ki]
young.woman-PST.ITR-FIN=DL
(creation myth, line 5)
(44) [sen-le-n-redo] go-T/A-ITR-WHEN/IF,
but [agu-li--a] bring-T/A-TR-FIN
(girl-snake story, lines 14, 16)
As well as tense aspect markers, some of the pronominal person clitics also undergo harmony.
Pronominal person clitics indicate the subject and can attach as clitics to the word immediately preceding the
verb.
(45) [cikna=la] what=1.DL.INCL, but [mai=le] food-1.DL.INCL
(creation myth, line 13)
(46) [aben=ben] 2.DL.2.DL, but [budi=bin] intelligence=2.DL
(creation myth, lines 30, 35)
There are also some examples of disharmonicity. For example, the plural morpheme /-ko/
sometimes harmonizes to [-ku] when that word contains a high vowel (48).
(47) [bare-ko] brother-PL'
[o-ko] fruit-PL
(48) [ilu-ku] meat-PL
[mai-ku] food-PL
However, this does not seem to be consistent. In the following examples, /-ko/ follows a high vowel
but does not harmonize.
(49) [somaw-budi-ko] intelligence-PL
(50) [dudlum-ko] pigeon-PL
It seems that the words that do not harmonize with /-ko/ are either compound words (49) or longer
words (50).
Although the pronominal person clitics above seem to harmonize, other clitics are not affected by
harmony. For example, the focus clitic /=do/ never seems to raise to [du] after a high vowel.
172 Anna Pucilowski

(51) [munu=do] not [munu=du] beginning-FOC
[muni=do] not [muni=du] hermit-FOC
Postpositions do not harmonize:
(52) [buru-te] forest-ABL
(53) [disum-re] world-LOC
The evidence suggests that vowel harmony in Ho is affected by morphological structure. At this
stage, we do not know all the morphemes that harmonize but we might predict that there is some reflection of
morphological closeness so that the more inflectional affixes such as tense/aspect and person markers will
harmonize, while the more derivational-like morphemes, the postpositions, as well as most clitics do not
harmonize.
Vowel harmony in Ho is a little studied area of research. We can say that the mid vowels harmonize
with the high vowels, while the low vowel /a/ raises to [e] or []. Given that vowel harmony is in the
direction of the high vowels, we see examples of both progressive and regressive harmony. Further research
will reveal all of the morphemes that normally undergo harmonization.
5.2 Rhythm and Intonation
As is the case with most Munda languages, Ho does not appear to be a tone language. Korku, another North
Munda language, is the only Munda language known to have tonal contrast (Zide 1966).
Very little is known about rhythm and intonation in Ho. The findings I present here for Ho must be
seen as preliminary. The following hypotheses are based on my subjective impression about which was the
prominent syllable in a given word.
Syllable weight seems to correlate with the prominent syllable in Ho. As discussed in section 4, we
can say that both CVV and CVC syllables count as heavy in Ho and these syllables are normally emphasized.
(54) ['sin.du.ri] vermillion
['kun.u] wooden post
['ce:.ra] pretty
[ra.cam] cut with scissors
Ho also has superheavy syllables and these always attract emphasis:
(55) [kon.ai] put branches into an earthen pot
When a word contains two or more syllables of equal weight, emphasis falls on the first syllable, as
in the following examples.
(56) ['mu.ni] hermit
['sa.du] holy man
['ra.ca] courtyard
['ca.ka.] snacks served with rice beer
['ge.e.ra] obstinate
This is less clear in words where we have two or more heavy syllables, e.g., CVC.CVC. In some
instances, it seems like the above rule applies, i.e., emphasis falls on the first syllable:
(57) ['sa.gar] hunt
['pam.pal] buttery, moth
However, in some instances of a homorganic nasal-stop sequence, the final syllable can also be
prominent:

(58) [ba.caw] save, salvation
[ban.dor] monkey
This may be related to the phenomenon of prenasalized stops (see section 2.1.1). The nasal, in these
cases, is acting as part of the onset of the second syllable rather than the coda of the first syllable. This
creates a heavier second syllable, which thereby attracts stress. Further evidence for this is the fact that
/bacaw/ is given as an alternative for /bancaw/ save, salvation (Deeney 2005:22) (both from Hindi bachana).
Although more research on rhythm is needed, we can hypothesize that syllable weight determines
word level prominence in Ho so that heavy syllables are emphasized. Where there are two or more syllables
of equal weight, the first syllable is stressed. Exceptions to this generalization are homorganic nasal-stop
sequences which occur word medially, creating a heavier following syllable which then carries stress. Future
research will compare word level prominence with rhythm and patterns of emphasis at the phrase and
sentential levels.
6 Conclusions
This paper has described the basic phonetics and phonology of the Chaibasa dialect of the Ho
language. We have seen that Ho has many intriguing features, including prenasalized voiced stops, which are
described here for the first time. Hos vowel system is also interesting. I have argued that vowel length is
indeed phonemic, based on minimal pairs and phonetic analysis. The vowel harmony system was described
in section 5.1 and is shown to be of a familiar type, based on vowel height, but sensitive to morphological
closeness.
More research is needed in many areas, including the prenasalized stops as well as details of the
suprasegmantal features of vowel harmony and stress assignment. Especially valuable to the study of Ho
phonetics would be sociolinguistic data so that we can see how Ho is changing, as well as the influence of
regional languages, such as Hindi and Oriya, which is presumed to be great.
References
Anderson, G. D. S. (ed.) 2008. The Munda Languages, New York: Routledge.
Anderson, G. D. S., Osada, T. & Harrison, K. D. 2008. Ho and the other Khewarian Languages, in G. D. S.
Anderson (ed.), The Munda Languages, New York: Routledge, pp. 195250.
Deeney, J. 2002. Ho Grammar and Vocabulary, (2
nd
edition). Ranchi: Xavier Ho Publications.
Deeney, J. 2005. Ho-English Dictonary, (2
nd
edition). Ranchi: Xavier Ho Publications.
Johnson, K. 2005. Acoustic and Auditory Phonetics, Malden, MA: Blackwell.
Lewis, M. Paul (ed.), 2009. Ethnologue: Languages of the World, Sixteenth edition. Dallas: SIL International.
Online version
<http://www.ethnologue.com>
Ohala, J. 1997. Aerodynamics of phonology, Proceedings of the 4th Seoul International Conference on
Linguistics [SICOL] 11-15 Aug 1997.
Osada, T. 2008. Mundari, in G. D. S. Anderson (ed.), The Munda Languages, New York: Routledge.
van der Hulst, H. & van de Weijer, J. 1995. Vowel Harmony, in J. Goldsmith (ed.), The Handbook of
Phonological Theory, Cambridge, MA: Blackwell, pp. 495534.
Zide, N. 1966. Korku low tone and proto-Korku-Kherwarian vowel system, in N. Zide (ed.), Studies in
Comparative Austroasiatic Linguistics, The Hague: Mouton, pp. 214229.
Zide, N. 1991. Lexicography of other languages of the Indian subcontinent: The Munda languages, in F. J.
Haumann, O. Reichmann, H. E. Wiegard L. Zgusta (eds), An International Encyclopedia of
Lexicography, Vol. 3, Berlin and New York: Walter de Gruyter, pp. 25332547.

Rau, Felix. 2011. Notes on Glottal Constriction in Gorum. In Sophana Srichampa and Paul Sidwell (eds.) Austroasiatic
Studies: papers from ICAAL4. Mon-Khmer Studies Journal Special Issue No. 2. Dallas, SIL International; Salaya,
Mahidol University; Canberra, Pacific Linguistics. pp.174-183.
174
Notes on Glottal Constriction in Gorum
Felix Rau
Leiden University
Introduction
*

In this paper I take a look at the phonology of glottal constriction in Gorum, a South Munda
language of the Austroasiatic stock spoken in the Eastern Ghats in India. The common denominator of the
phenomena I focus on in the following is glottalization, i.e. stricture at glottal level, and can be represented
by the feature [+constricted glottis]. In the present study I emphasize the role of the syllable and other
prosodic units. Through this focus I take up one aspect of an older approach by Aze (1971, 1974), which is
now generally ignored. Additionally, I will highlight the comparative and historical implications of this
analysis.
The three phenomena I am interested in are the following: the glottal stop //, as in (1), a series of
preglottalized voiced obstruents /b, d, , (g)/, such as /d/ in (2), and vowels with creaky voice articulation,
i.e. /a o /, as /a / in (3).
(1) a [a] water
(2) ad [ad] for
(3) a l [a l] husking pit
The glottal stop and creaky voice are purely glottal phenomena and differ from one another mainly
in the degree of glottal constriction and its timing relative to the vowel. The glottal stop is a complete
obstruction of the airflow at glottal level and is perceptionally clearly delimited from the adjacent vowel.
Creaky voice involves a lesser degree of glottal constriction and extends over the whole duration of the
vowel and a following sonorant in the rhyme, if present. Perceptionally, creaky voice is a property of the
vowel, as it cannot be separated from it. The glottalization in preglottalized obstruents, on the other hand, is
part of a complex phenomenon and occurs at the boundary between vowel and obstruent, parallel to the oral
closing gesture. The glottalization is here only one aspect of the phoneme.
Most other Munda languages only have the glottal stop and the (pre-)glottalized obstruents, although
Juray (Zide 1982) and Sora (Donegan p.c.) also have creaky phonation as a variant of the glottal stop. In most
Munda languages, the (pre-)glottalized obstruents are considered allophones of the non-glottalized obstruents,
as they occur only in syllable-, stem-, or morpheme-final position. This is the case, for example, in Santali
(Ghosh 2008) and Kharia (Peterson 2008). Gorum, however, seems to be unique among Munda languages in
having all three phenomena. Also, as I will argue, these phenomena are phonemic and involve one feature of
glottalization, whose proper domain is the syllable.

* I would like to thank all ICAAL 4 participants and especially Patricia Donegan and Gerard Diffloth for their
helpful comments and stimulating discussions. Additionally I am also indebted to Juliette Huber for commenting on
earlier versions of this paper. Some of the data used in this paper were collected on field trips partially funded by the
Leiden University Fund.
Felix Rau 175

Previous Accounts
Aze (1971, 1974) and Zide (1963, 1982) analyze the status of these three phenomena in very
different, in fact incompatible, ways. Aze (1971, 1974) subsumes the three under a single prosodic phoneme,
an analysis that allows him to reduce Gorum syllable structure to (C)V(N) and (C)V(N), where //
represents glottalization of the syllable. However, his analysis cannot distinguish between creaky voice
phonation and the glottal stop. Furthermore, it fails to distinguish a combination of either one of these with a
nasal from the homorganic preglottalized obstruent. As such, Azes analysis cannot account for a set of
minimal pairs and hence has to be regarded as insufficient.
Zide (1963, 1982), on the other hand, recognizes three distinct phenomena, assuming that creaky
voice and the glottal stop are two distinct phonemic segments, while preglottalized obstruents are a non-
phonemic variant of voiced obstruents. The treatment of creaky voice as a segment is phonologically
unfortunate. Also, while the glottalization of obstruents might very well have been a non-phonemic process
in earlier stages of Gorum, this analysis seems not to be appropriate to account for the present state, nor does
it seem to be psychologically adequate.
In my own analysis, all three glottal phenomena involve one feature [+constricted glottis] or
glottalization, which is connected to the syllable. In so far, thus, I follow Aze. However, in my view all three
are distinct and phonemic, so that in this respect I am more in agreement with Zide. Historically, I believe she
is right in assuming that the preglottalized obstruents are not phonemic. However, synchronic evidence
shows that they have become so, due to the heavy influx of Indo-Aryan loan vocabulary, so that
preglottalized obstruents contrast with non-glottalized ones in coda-/stem-final position.
The Phonemic Status of Glottal Phenomena
The fundamental differences in the previous accounts illustrate the difficult phonemic status of the
three phenomena. It is in fact difficult to demonstrate their distinctiveness, since they are largely confined to
mutually exclusive contexts; also, the phonetic differences can be very subtle.
Each of the three phenomena contrasts with its absence, i.e. [+constricted glottis] contrasts with
[constricted glottis]. The examples in (4)-(9) demonstrate this. The difference between (8) and (9), however,
is a secondary one between native and loan vocabulary, as (9) is a loan from Telugu, probably via Desia
Oriya. Synchronically, it is nevertheless real and does not seems to be different from the other two contrasts.
Examples (8) and (9) are especially relevant in the light of Zides claim that preglottalized obstruents are not
phonemic.
107

(4) a l [a l] husking pit
(5) al [al] to thatch
(6) a [a] water
(7) a [a] to do
(8) abu [abu] close-INF.TR
(9) abu [abu] money
Having established the contrastivity of presence vs. absence of glottal stricture in all three types, let
me now come to the contrast between the three phenomena. Finding minimal pairs here is rather more
complicated. The main reason for this is that the preglottalized obstruents are, by their very nature as oral
obstruents, quite distinct from the purely glottal phenomena creaky voice and glottal stop. The pairs (10) and

107 The distinctiveness can still be doubted in cases such as (8) and (9). These two can be seen as structurally
different i.e. by construing the syllable structure of (8) as /ab.u/ in contrast to /a.bu/ for (9). This, however, does not
invalidate the fact that the distribution of both types of phonemes widely overlaps, so that no general complementary
distribution can arise.
176 Notes on Glottal Constriction in Gorum

(11) as well as (12) and (13) come as close to minimal pairs as one can get. All four feature glottal
constriction combined with an alveolar closure. In case of /Vn/ in (11) and /V n/ (12), this involves nasality
in the form of a nasal segment, in this case the nasal stop /n/, and in (10) and (13) a nasal release of a single
complex phoneme /d/ [Vd].
(10) ad [ad] for
(11) kina-n [an] at the river (river-LOC)
(12) a-nn [nn ] if done (do-TOP)
(13) abged [ged] to ignite
These combinations of glottal constriction, alveolar oral closure and nasality differ slightly in their
phonetics. In the glottal stop plus nasal combination /Vn/ in (11), the vowel preceding the glottal stop as
well as the nasal following it are minimally affected by the complete glottal stricture. This results in a sound
event best transcribed as [an]. The combination of a creaky vowel with a nasal as in (12), shows a glottal
constriction that stretches over the whole articulation of the vowel and extends into the following sonorant.
This sound event is best represented by the transcript [n ]. The preglottalized obstruents in (10) and (13) are
more complicated in their articulation. The vowel starts in modal voice phonation. The glottal stricture starts
parallel to the oral gesture and is clearly audible before the oral closure is complete. The stop is afterwards
released in nasal plosion. This sound event may be transcribed as [ad], where [] is intended to represent the
temporally restricted glottal constriction at the end of the vowel.
The three types of glottal phenomena differ primarily in the relative timing of the gestures involved,
i.e. glottal constriction, oral closure and velic opening, as well as in the degree of glottal stricture, i.e.
complete closure in the case of // and partial constriction in the case of creaky voice and preglottalization.
Speaker judgments vary in their rigidity: while speakers regard a replacement of /d/ by /n/ as a
clear audible mistake, they are less clear about the relationship between /d/ and /V n/.
All three types of phenomena are consonantal. Interestingly, even creaky voice, which from a
phonetic point of view seems to be a quality of the vowel, behaves in some respects like a consonantal
phoneme. The best evidence for this comes from echo word formation. Echo words are a phonologically
altered repetition of a word. One strategy used in Gorum is vowel replacement. In this formation process, the
consonantal skeleton of the base word is maintained, while the vowel or vowels are replaced.
(14) ali liquor
(15) ula liquor (echo word)
(16) gaga cooked rice
(17) gigi cooked rice (echo word)
(18) gmar winnowing
(19) gmir winnowing (echo word)
In this process, creaky voice figures as part of the consonantal skeleton, just like the other elements
with glottal constriction. Thus in the case of examples (18) and (19), the consonantal skeleton of the word is
gV .mVr, with creaky voice a part of it, as shown by the fact that the creaky voice in the first syllable is
unaffected when the original vowel pattern u-a is replaced in the echo word by i-i.
In summary, the three types of glottalized phonemes glottal stop, preglottalized obstruents and
creaky voice have been shown to be distinct and phonemic in Gorum. They behave as a class and can be
categorized by the feature [+constricted glottis]. Evidence from echo word formation suggests that all three
should be grouped with consonants. This is, however, all that can be gained from a point of view centered on
Felix Rau 177

the segment. All other relevant data is syllable- or even stem-related and will be discussed in the following
sections.
I can now contrast my view of the phonemic status of the glottal phenomena with the two previous
accounts by Aze and Zide. For Aze (1971, 1974) every occurrence of glottalization is distinctive and
glottalization is a prosodic property of the syllable. However, he recognizes only one phonemic process of
glottalization, so that in his analysis creaky voice phonation and the glottal stop are identical. Additionally,
he regards the phenomenon here called preglottalized obstruents not as an obstruent phoneme, but as a
combination of a nasal coda with prosodic glottalization. By this he can reduce all syllable structures in
Gorum to four types: CV, CV, CVN, CVN. Zide, on the other hand, only recognizes segments and
distinguishes a phonemic glottal stop from a phonemic creaky voice. Preglottalized obstruents in her analysis
are obstruents. However, they are not glottalized phonemically, but allophones of the non-glottalized voiced
obstruents.
Since Aze recognizes glottalization only on syllable level, the three approaches can best be
compared by contrasting the possible syllable types that result from the approaches. Note how several
syllable types are conflated in Azes analysis, while Zides analysis basically yields the same results as mine.
However, she treats creaky voice as a segment /H/ and syllables with preglottalized obtruents are not
phonologically glottalized in her view.
Rau Aze Zide
CV CV CVH
CV CV CV
CVn CVn CVn
CV N CVn CVHN
CVO CVn CVO (not glottalized)
Table 1: Comparison of the different approaches
The Syllable
The syllable is of particular importance for the phonology of glottal constriction in Gorum, since
restrictions on number and placement operate on syllable-level. Glottal constriction is restricted to the rhyme
and only one occurrence is permitted per syllable.
Syllable Structure
Gorum has a maximal CVC/CV C syllable structure. Three exceptions occur: The complex onset /r/
occurs in one native word, two lexemes have a complex coda /k/, and some word forms possess a coda /n/.
This last cluster can only occur as a result of the affixation of the locative marker -n and is very rare; an
example can be found in (11) above.
The feature [+constricted glottis] is confined to the rhyme. All three types of phenomena occur with
the nucleus or after it.
108
Additional phonetic evidence comes from the articulation of creaky voice. In words
such as a l [a
l] husking pit in (20) (repeated from example 4), the glottal stricture extends into a sonorant in
coda position. However, the creaky phonation never extends into a sonorant in the onset of the following
syllable. Thus the /m/ in a ma [a .ma] before is not affected by the creaky phonation of the first syllable.
(20) a l [a
l] husking pit
(21) a ma [a .ma] before

108 According to my analysis, this restriction applies to all syllables in all situations. Hence, words such as
/ab.u/ close-INF.TR (from ab to close) violate the maximal onset principle. Nevertheless, the phonetics and the
syllabication by speakers in slow speech conrm this interpretation. An alternative analysis with the syllable structure
/a.bu/ would be possible. Under this interpretation, the restriction on glottal constriction would only hold for the stems
in the lexicon and not for actual word forms.

In addition to the positional restriction, the three glottalized elements cannot combine with each
other to form a cluster. The constraints on the distribution of these sounds lead to the situation that
[+constricted glottis] can occur only once in a syllable. The positional and combinatorical constraints of the
three phenomena result in the syllable patterns represented in Table 2:
109
They are grouped here into four
groups. This first group comprises open syllables together with syllables with a liquid or nasal in coda
position. The second group consists of the same syllable types, but with creaky voice phonation. The third
and fourth groups are syllables with a glottal stop or a preglottalized obstruent in coda position. These four
groups are relevant for the following discussion of syllable weight.

(C)V (C)V (C)V (C)VO
(C)V(j/r/l) (C)V (j/r/l)
(C)VN (C)V N
Table 2: Syllable structures of Gorum
Syllable Weight
Syllable weight seems to be intimately connected to glottalization. (C)V syllables involving glottal
stops and (C)VO syllables with preglottalized obstruents are heavy. With creaky voice, the situation is more
complex. It can occur with open (C)V syllables, (C)V (j/r/l) syllables with liquids in the coda, as well as
(C)V N syllables with nasals in coda position. Without creaky voice, open syllables and syllables with glides
are light syllables, but with creaky voice they are heavy. Non-glottalized syllables with nasals are ambiguous
with respect to syllable weight.
Evidence for the relation between syllable weight and creaky voice comes from nominals. Anderson
and Zide (2001) propose a bimoraic constraint on nominals in Proto-Munda; a similar constraint seems to be
at work in Gorum, where most nouns are disyllabic. Of the monosyllabic nouns, most have the clearly
bimoraic form (C)V or (C)VO. There are no monosyllabic nominals with light syllables such as CV or
CV(j/r/l). There is, however, a small group of nouns of the form (C)V (j/r/l). Perhaps the most telling of them
is the following pair:
(22) sur to hunt
(23) sr hunting / a hunt
While the verb in (22) has the form CVr, the corresponding noun in (23) contains an additional
creaky voice and has thus the form CV r. This is particularly interesting given that the bimoraic constraint
only applies to nouns and not to verbs. Since the glottalization seems not to be part of the root, its presence is
either the source or an effect of the second mora.
Thus, syllable weight in Gorum is closely connected to the presence of glottalization. All syllables
containing [+constricted glottis] are heavy and can be considered bimoraic, although it is not evident in all
cases which is the cause and which is the effect.
110

109 Additionally, there is a very small number of exceptional native lexemes such as /lup/ [lup] big
which end in a voiceless obstruent. These lexemes deviate from the patterns presented here. These voiceless obstruents
also do not involve glottalization, but their laryngeal component is an aspirated release.
110 The status of syllables with a nasal in coda position is ambiguous, as pointed out above. Monosyllabic
nominals with this coda type tend to be (C)V N. However, in contrast to clearly light syllables with a liquid phoneme in
coda position, there are a few exceptions. The two pronouns mi I and ma you(SG) as well as the singular noun za
bone do not contain glottalization and either violate the bimoraic constraint or are bimoraic without any interaction with
glottalization.
Felix Rau 179

The Phonological Stem
The phonological stem is the next higher level of prosodic organization relevant to the phonology of
glottalization in Gorum. There is a constraint on the number of glottalized syllables operating on stem-level.
On the other hand, the stem is also crucial for the placement of glottalization in affectedness marking.
The Glottal Constraint
On the level of the phonological stem, there is a restriction on the number of syllables with
glottalization that can occur. In general, only a single glottal element is allowed in a stem. Thus in
reduplication of roots containing glottalization, the glottalized element is lost in the reduplicant. This is
illustrated in examples (24) through (26); the reduplicant in (24) is a faithful copy of the stem, while in (25)
and (26) the segment in the coda which contains the glottalization is lost in the reduplicant.
(24) zum to eat zumzum
(25) ga to eat gaga
(26) gad to cut gagad
This constraint on glottal elements is also at work with the causative prefix, which takes the form
ab- with roots containing glottalization and ab-
111
with roots without the constricted glottis feature. In
contrast to the reduplicant in (26), where the glottalized coda is lost, the coda segment /b/ or /b/ is present
in both (27) and (28), but the presence of glottalization depends on the nature of the root.
(27) ab + so causative + to learn ab-so
(28) ab + su causative + to fall ab-su
The three types of glottal phenomena form a class which can be interpreted as natural if we assume
that all three involve the same feature [+constricted glottis]. Furthermore, the stem is the relevant domain
here. This can be seen from example (29), which contains two glottal elements. However, the preglottalised
obstruent /d/ is part of the stem imad to sleep, while the glottal stop // marks affectedness on the non-
past suffix -tu. This shows that the constraint does not extend over a stem boundary.
(29) imad#-tu sleep#-NPST:AFF
The affectedness marker is discussed in more detail in the following section.
Affectedness Marking
Affectedness marking is another part of Gorum grammar in which glottalization interacts with
syllable structure and the phonological stem. This morpheme marks the medium voice of a verb. It is
obligatory with one class of verbs, while it has detransitivizing effects on other verb classes.
The morpheme takes two forms, depending on the syllable structure it encounters. In those cases
where it combines with an open syllable, it has the form of a glottal stop, resulting in a (C)V syllable. If the
coda position is occupied by a nasal, affectedness marking takes the form of creaky voice phonation,
resulting in a (C)V N syllable.
112
In either case, affectedness marking takes the form of glottalization and is
thus solely an instantiation of [+constricted glottis].
113

111 The glottalized form of the prex could be either ab- or a b-. The phonetic evidence is not unequivocal, yet
the preglottalized obstruent is the most likely solution.
112 Other forms do not occur, as verbal suxes only have CV or CVN syllable structures.
113 This is a clear case of allomorphy. If the aectedness marking is considered in isolation, it looks very much
as if the glottal stop and creaky voice were allophones conditioned by the prosodic structure. This seems to have
inuenced Azes conception of the phonological status of the three types of glottal phenomena.

The morphological behaviour of the affectedness morpheme is unique in the grammar of Gorum. Its
placement is not relative to other morphemes, but solely prosodic. The vital notions determining the
placement are the right stem boundary and syllable structure; it occurs in the first rhyme following the stem
regardless of any other aspect of the morphological structure of the verb. Due to the general restriction on
stem boundaries in Gorum to coincide with a syllable boundary, the syllable following the stem is the place
of affectedness marking. This is schematically represented in (30). (31) to (34) give concrete examples of the
placement relative only to the stem boundary, as well as of the allomorphy depending on the syllable
structure. They also show that morpheme order and other morphological aspects are irrelevant.
(30) STEM#
AFF
(31) (ne)
-(ko)
(ko)
#-(tu)
I will sit
1sA-sit-NPST:AFF
(32) (or)
-(gi)
#-(n-a j)
he/she/it is not visible

NEG:NPST-see-INF.INTR-CISL:AFF
(33) (u)
(ku)
#-(r-j)
they were
be-PST-3pS:AFF
(34) (u)
(k#-)
-(aj)
he/she/it is to me
be-1sP:AFF-CISL
Example (34) is exceptional: the right boundary of the morphological stem in Gorum generally
corresponds to a syllable boundary. However, in the case of the irregular verb uku to be, the stem
boundary does not coincide with a syllable boundary, but the affectedness morpheme is nevertheless
positioned directly after the boundary in the rhyme of this syllable.
Stem Patterns
In addition to the rules discussed above, there are some strong statistical tendencies in the
distribution of the three types of glottal phenomena in stems. The examples (35) to (42) give an impression of
these patterns.
(35) ga to eat
(36) seb to chop
(37) bl to be drunk
(38) tu.pad to thresh
(39) kin.a river
(40) go .tu cloth
(41) a.an.ad door
(42) bi.o .gi tomorrow
As can be seen, there is a strong tendency for the glottal stop // and the preglottalized obstruents
/O/ to occur in the last syllable of a stem and, as such, at its right boundary, while creaky voice /V / prefers
the penultimate syllable. In monosyllabic stems this tendency is neutralized, so that no complementary
distribution arises. These patterns result in the following stem structures:

Felix Rau 181

monosyllabic stems disyllabic stems trisyllabic stems
CV CVC.CV CVC.CVC.CV
CVO CVC.CVO CVC.CVC.CVO
CV C CV C.CVC CVC.CV C.CVC
Table 3: Distribution of glottalization in word stems
In a small number of stems this pattern is broken, but most of these cases seem to be frozen
compounds. An example is the word mita n today, in which creaky voice occurs in the last syllable.
However, this lexeme, which speakers today view as a simplex, can be analyzed as a compound of the word
mit day and the synchronically unattested demonstrative *a n, probably meaning this day. This putative
demonstrative can be connected to the still used a t that day, parallel to the pair o n hither and o t thither.
A peculiar pattern occurs with some Telugu or Oriya loan words which have a CV.CV structure in
the source language. In Gorum, their first syllable became glottalized, so that the resulting prosodic structure
is CV .CV. This pattern is not productive anymore, but at some point in the history of Gorum it must have
been. The following two words are good representatives of this group.
(43) o pa leaf bowl from Telugu doppa
(44) ka u bangles from Desia Oriya kau
The motivation for the presence of creaky voice in the first syllable of these words is unclear. None
of the source languages has creaky voice as a sound phenomenon, let alone as a phoneme. Some of these
lexemes, such as (43), contain a geminate in the source language, a phonological structure that does not exist
in Gorum. However, since not all of these lexemes originally contained geminates, this cannot be the
defining condition for the phenomenon. Whatever the motivation for the creaky voice in these words, they
show that under some conditions glottalization could come into existence in lexemes that originally did not
have it. This poses some problems for the historical reconstruction of this aspect of Gorum phonology. I will
address this issue in the following section.
Historical Significance
The previous discussion especially the occurrence of creaky voice in loan words raises some
problems for the historical treatment of glottalization in Gorum. Of the three glottal phenomena, two are
comparatively well understood. The glottal stop is present all over the Austroasiatic family and preglottalized
obstruents are found in virtually every Munda language. Also, the history of both seems to be relatively
straightforward.
Creaky voice, however, is intriguing: it occurs in several Austroasiatic languages, but its history is
still nebulous. Gorum is the only Munda language in which it is phonemic. Nevertheless, it has been claimed
to be reconstructable for Proto-Munda (Zide 1976 as well as Zide and Zide 1987). Diffloth (1989) argues that
it goes back to Proto-Austroasiatic. Indeed the evidence from all over the language family is quite suggestive,
yet the history of this phoneme is still not well understood. This may be due to a variety of reasons. In the
Munda languages, it appears that several developments, besides segmental sound changes, have obscured the
picture.
The most frequent occurrence of creaky voice in Gorum is probably the affectedness morpheme.
This morpheme has no known direct correspondences in other Munda languages, so its history is difficult to
determine. Its allomorphy with a creaky voice and a glottal stop allomorph considerably complicates the
historical phonology of this morpheme: given the phonological similarity between the two allomorphs, every
putative comparative evidence could pertain to either one of the two glottal phonemes.
Further difficulties for a comparative approach stem from the fact that in nominals such as sr hunt
(example 23), the creaky voice seems to be connected with the second mora and might very well be a

derivational device. In other lexemes such as a su house the creaky voice does not seem to be part of
the root either. In this particular case, internal and comparative evidence suggests that the root is su, while
a - seems to be a derivational prefix. Thus in some cases the creaky voice cannot be established as part of the
root and its emergence at some point in history is not understood yet.
114

In addition to those cases were creaky voice cannot be ascertained to be part of the etymon, there are
also words where it cannot possibly be part of the original root. These loan words with a CV.CV-structure are
evidence that at some point in the history of Gorum a productive pattern existed that gave rise to creaky voice
in these lexemes. As long as the motivation for its emergence in these words is not understood, this poses a
severe problem for the historical treatment of creaky voice.
Perhaps the most fruitful starting point for a comparative approach would focus on words in which
creaky voice occurs in the root and no known pattern motivates its emergence. To my knowledge the
monosyllabic verbs in (45)-(48) are the best candidates for this. The other four lexemes are also likely
candidates. The only caveat is that la ki has a CVCV structure, but in contrast to the loans discussed above it
is neither a loan word nor a noun. However, this list cannot resolve the fact that prosodic structure and word
phonological processes seem to be the key to the understanding of this phenomenon.
(45) bl to be drunk
(46) l to ripen
(47) m to live
(48) n to perform a burial ritual
(49) a ma before
(50) a gaj when
(51) la ki later
(52) bio gi tomorrow
There is, I think, another option which should be considered: the phonemic creaky voice in Gorum
could be the result of a split in this language, rather than a remnant of an old Proto-Munda phoneme. The
tendency for a complementary distribution of creaky voice and glottal stop in stems and the allomorphy of
affectedness marking could be interpreted in that direction. The current status of creaky voice in Sora and
Juray would then be similar to the historical situation in Gorum.
On the other hand, Diffloths evidence strongly suggests that creaky voice is an old feature in the
Austroasiatic family. However, the historical development, at least on the Munda side of the family, is so
heavily obscured that historical reconstruction has to proceed with care.
Conclusions
There are three phonemic types of glottalization in Gorum: the glottal stop, preglottalized obstruents,
and creaky voice. These types are forms of one general phenomenon glottalization that can be represented as
[+constricted glottis]. Its phonology is best understood in terms of syllable structure. Also, interactions with
other elements, restrictions and regularities have to be captured on the even higher level of the phonological
stem.

114 I am not convinced by the connection made by Zide and Zide (1987) between the loss of some instances of
/s/ in Sora-Juray-Gorum and creaky voice. Since we do not understand the history of creaky voice and there are no
systematic, if any, reexes of it in other Munda languages (cf. Zide 1982, p. 367), it seems problematic to posit its
existence in some lexemes at some earlier stage solely to explain another unexplained phenomenon.
Felix Rau 183

The history of these phonemes is still poorly understood and in my opinion, a better understanding
of prosodic structures in Munda languages, and especially the phonology of glottal constriction, is needed to
understand the diachrony of these phenomena.
References
Anderson, G.D.S. and Zide, N.H. 2001. Issues in Proto-Munda and Proto-Austroasiatic Nominal Derivation:
The Bimoraic Constraint, In: Macken, M.A. (ed.) Papers from the 10th Annual Meeting of the
Southeast Asian Linguistic Society. Arizona State University, Tempe. 5574
Aze, R., 1971. Parengi (Gorum) Phonemic Summary. Summer Institute of Linguistics and Deccan College,
Poona.
Aze, R., 1974. The status of glottalization in Parengi (Gorum). Proceedings of All India Conference of
Linguistics, 1972, 8495.
Diffloth, G., 1989. Proto-Austroasiatic creaky voice. Mon-Khmer Studies 15, 139154.
Ghosh, A. 2008. Santali. In: Anderson, G. D. S. (ed.) The Munda Languages. Routledge Language Family
Series. Routledge, Oxon. 11-98.
Peterson, P. 2008. Kharia. In: Anderson, G. D.S. (ed.) The Munda Languages. Routledge Language Family
Series. Routledge, Oxon. 434-507.
Zide, A. R. K., 1963. Gorum (Parengi) Phonology, Unpublished M.A. thesis, University of Pennsylvania.
Zide, A. R. K., 1966. Some problems in the generation of Gorum obstruents and related glides. Uuniversity
of Chicago (mimeo).
Zide, N. H., 1976. 3 and 4 in South Munda, Linguistics 174:89-98.
Zide, A. R. K., 1982. A reconstruction of Sora-Juray-Gorum phonology. PhD thesis, University of Chicago.
Zide, A. R. K., Zide, N. H., 1987. A KM laryngeal as a conditioning factor for s-loss in Sora-Juray-Gorum.
In: Cardona, G., Zide, N. H. (Eds.), Festschrift for Henry Hoenigswald: On the occasion of his
seventieth birthday. Narr, Tbingen, 417431.

Subbarao, Karumuri V. Gracious M. Temsen. 2011. Comitative PP as Head in Externally-Headed Relative Clauses
in Khasi. In Sophana Srichampa and Paul Sidwell (eds.) Austroasiatic Studies: papers from ICAAL4. Mon-Khmer
Studies Journal Special Issue No. 2. Dallas, SIL International; Salaya, Mahidol University; Canberra, Pacific Linguistics.
pp.184-201.
Copyright vested in the authors.
184
Comitative PP as Head in Externally-Headed Relative Clauses in Khasi
115

Karumuri V. Subbarao Gracious M. Temsen
University of Hyderbad University of Hyderbad
Introduction
This paper attempts to discuss the case of the comitative PP as head in externally-headed relative
clauses (EHRCs) (also labeled as: gap relatives) in Khasi, a Mon-Khmer language. While a sentential relative
clause and a gap relative (EHRC) are both permitted with subject, direct object, indirect object, and all
oblique objects (instrumental, ablative, locative PP) as head, the comitative PP is unique in not permitting a
sentential, as well as gap relative, as head with the interpretation of accompaniment. A similar phenomenon
is observed in all other South Asian languages and some other non-South Asian languages with regard to the
comitative PP as head in gap relatives.
While there is considerable variation with regard to the positions that are accessible in the Noun
Phrase Accessibility Hierarchy (NPAH) of Keenan and Comrie (1977), in many South Asian languages of
the four language families (Indo-Aryan, Dravidian, Austro-Asiatic (Mon-Khmer, Munda) and Tibeto-
Burman), the modification of the comitative PP in gap relatives is not permitted in almost all South Asian
languages. In the few languages in which it is permitted, specific syntactic criteria have to be fulfilled to
enable the comitative PP to head the gap relative. In this paper we attempt to propose a condition which we
shall label as the Thematic Eligibility Condition (TEC), and wish to demonstrate that only after fulfilling
specific syntactic criteria, the requirement of TEC is satisfied to enable a comitative PP to head a relative or
gap relative clause.
The processes that enable a comitative PP to head an EHRC or an IHRC include:
(i) Comitative Adposition Incorporation in the embedded verb; as in Mizo, Hmar, Mising (Tibeto-
Burman); or
(ii) the overt occurrence of the postposition with the internal head of the comitative PP in IHRCs; as in
Tenydie, Konyak and Sangtam (Tibeto-Burman); or
(iii) the occurrence of the verbal reciprocal that functions as a group marker together with the occurrence
of an adverb with the interpretation of together in the embedded verb; as in Khasi (Mon Khmer)
and Manipuri (Tibeto-Burman).
Thus, either some elements are added to the embedded predicate of the gap relative, or some
syntactic process such as Adposition Incorporation takes place, or the internal head of an IHRC must have an
overt postposition indicating accompaniment as in some Tibeto-Burman languages in the few languages that
permit a comitative PP to head an EHRC or an IHRC.

115
We are thankful to Alice Davison, Rajesh Bhatt, Hans H. Hock, George van Driem and Juanita War for their
suggestions. Thanks are also due to Matt Shibatani (for Japanese and Korean), Tatiana Oranskia (for German & Russian),
Ludmila Khoklova and Boris Zakharyn (for Russian) for providing either data or information concerning the occurrence
of the comitative PP in relative clauses. Thanks are also due to Paul Sidwell for his continuous support in many ways.
Comitative PP as Head in Khasi 185

In this paper we provide data that demonstrate that the process mentioned in (iii) above is found in
Khasi in EHRCS and we provide arguments to show why such markers are included in the embedded verb.
We shall show that comitative PPs (either Prepositional or Postpositional) that are thematically
linked to the embedded predicate freely permit the modification of the comitative PP. We argue that a
thematic relation has to be established between the predicate of the embedded clause and the head of the
EHRC/IHRC, in terms of either an overt case marker with the head, or some marker in the embedded verb to
indicate accompaniment. In Khasi, we show that such a thematic relationship is established by the
incorporation of the adverb -la-together, and the verbal reciprocal -ya-in the embedded verb. The verbal
reciprocal -ya-in Khasi also functions as a group marker, a hortative marker, collaborative effort marker (cf.
Subbarao ms.). We also show that in case the embedded predicate is inherently reciprocal, the adverb -la-
together is not needed, as a reciprocal predicate necessarily implies a combined activity by two or more
than two participants.
2. A few facts about Khasi
Khasi, a Mon-Khmer language, is the only non-verb-final language in the South Asian
subcontinent
116
. It has predominantly verb-medial structures, though there is a set of structures in which the
verb obligatorily occurs in the initial position (see Temsen, 2007).
Conforming to its non-verb-final structure, Khasi has prepositions and the auxiliary verb invariably
occurs to the left of the verb. The complementizer ba occurs to the left of the embedded complement clause.
Khasi has subject-verb agreement and all adjectives exhibit agreement with the modifying noun. Personal
pronouns such as u-he, ka-she, and ki-they are homophonous with agreement markers on the verb, the
noun and the adjective as markers indicating gender. Khasi is also a null pronominal language. It has both
verbal and nominal reciprocals. The verbal reciprocal performs several other functions such as a group
marker, a hortative marker, and a collaborative effort marker (cem) (see Subbarao, ms.). Khasi has full
sentential relative clauses (The English wh-type) and Externally-Headed Relative Clauses (EHRCs), also
called as gap relatives, and conforming to its non-verb-final pattern, it does not have Internally-Headed
Relative Clauses (IHRCs). The marker ba which occurs with adjectives also functions like an adjectivalizer
in gap relatives as well as a relative pronoun.
3. Relative Clauses and EHRCs in Khasi
3.1. Relative Clauses in Khasi
The relative pronoun is ba, it occurs to the left of the embedded clause. We provide below two
examples of Oblique Object modification. Sentence (1) is an example of a sentential relative clause. The head
is a locative PP and the locative object m
y
d table is modified.
Locative PP as head Relative clause
1. ka-m
y
d
1
ha-ka ba u-lam u-b ya-ka-kt t
1
ka-la-kd
y
a
f.s-table loc-3f.s which m. s-Lam 3m.s-put acc-f.s-book 3f.s-perf-break
The table on which Lam put the book is broken.
Sentence (2) is an example of ablative object modification in which the ablative object jaka place
is modified.

116
Pnar, Mnar, Jowai and Rymbai (Mon-Khmer languages) are also verb-medial like Khasi. See Bareh (2007)
for Jowai and Rymbai.
186 Karumuri Subbarao, Gracious Temsen

Ablative PP as Head -Relative clause
2. ka-jaka
1
na-ka ba u-lam u-wan t
1
ka-l ka-ba- ji
f.s-place loc-3f.s which m. s-Lam 3m.s-come 3f.s-be 3f.s-adjr- far
The place which Lam came from is far.
In (1) and (2) the relativized noun phrase (in bold in (1) and (2)) moves leftward to the initial
position, namely, to Spec CP of the main clause. An important fact about the relative clause in all South
Asian Languages (SALs) is that the relativized NP is generally fronted, if it is a non-subject (DO, IO, or OO),
just as in English and French. The position from which the NP is fronted is indicated by the trace t
i
, which is
coindexed with the relativized NP in (1) and (2).
3. 2. EHRCs in Khasi
An Externally-headed (Gap) relative that corresponds to (1) is given in (3). The embedded clause in
an EHRC functions like an adjectival modifier with ba functioning as an adjectivalizer.
Locative pp modified-ehrc ( gap relative)
3. ka-m
y
d
i
[
AP
ba
117
- u-lam u-b ya-ka-kt
i AP
] ka-la-kd
y
a
f.s-table adjr- s. m-Lam 3m.s-put acc-f.s-book 3f.s-perf-break
The table Lam put the book on is broken.
In the Gap Relative in (3) above: (i) there is a gap (a null operator) in the embedded clause
(indicated by ), as the identical noun phrase does not overtly occur; (ii) the locative preposition indicating
the grammatical relation between the head of the gap relative and the embedded predicate does not occur, as
EHRCs universally do not carry an adposition (preposition or postposition) with the head that is relativized;
and (iii) the embedded clause occurs to the right of the head noun as modifiers in Khasi always occur to the
right of the head they modify.
For each position on the Noun Phrase Accessibility Hierarchy (NPAH)
118
of Keenan and Comrie
(1977) there is a sentential relative clause as well as a corresponding EHRC available in Khasi, except for the
Comitative PP unless some elements are incorporated in the embedded verb. The positions that are
relativizable in Khasi are: the Subject, the Direct Object, the Indirect Object, the Object of the Genitive, the
Oblique Object, which includes the Locative, the Ablative, and the Instrumental PPs and so on.
Formation of EHRCs with the Comitative PP as Head
This section provides a detailed description on the relativization of the Comitative PP as Head
indicating accompaniment. In sentence (4) below, we first provide an example of the comitative PP in a
simple clause.
4. u-lam u-wan bad ka-wei ka-khinna a-ka-jikhawai
m.s-Lam 3m. s-come com 3f.s-one f.s-child to-f. s-part y
Lam came to the party with a girl.

117
Adjectives in Khasi too carry the marker ba. Thus, the entire following phrase functions like an adjectival
phrase (AP):
ba- u-lam u-b ya- ka-kt
adjr- m,s-Lam- 3sm-put acc-3f,s-book
118
The NPAH constraints proposed in Keenan and Comrie (1977) are:
1. A language must be able to relativize subjects.
2. The strategy of relativization must apply to a continuous segment of the NPAH-scale.
3. Strategies that apply at one point of the NPAH-scale may in principle cease at any point on the scale.

In sentence (5) below, the group marker -ya-and the adverb -la-together occur with the verb. The
occurrence of these two markers together imparts the specific interpretation that the act of arriving of Lam
and the girl at the party happened at the same point of time, while sentence (4) does not necessarily have the
implication that Lam and the girl came together.
5. u-lam u-ya
119
-wan-la bad ka-wei ka-khinna a-ka-jikhawai
m.s-Lam 3m. s-gm-come-together com 3f.s-one f.s-child to-f. s-part y
Lam came to the party together with a girl.
Sentences (6a & 6b) and (7) below provide evidence in support of our claim. Sentences in (6) are
grammatical, while sentence (7) is not, as it is a contradiction. Thus, when two individuals perform an action
together towards achieving a goal with a common purpose, the adverb -la-together and the verbal
reciprocal -ya-which also functions as a group marker obligatorily occur with the verb.
6a. u-lam bad ka-wei ka-khinna ki-ya-wan
m.s-Lam com 3f.s-one f.s-child pl-gm-come
a-ka-jikhawai ta-ba ki-khlm-ya-wan-la
to-f. s-part y only-t hat pl-neg-gm-come -toget her
Lam came to the party with a girl but they did not come together.
6b. u-lam u-wan a-ka-jikhawai bad ka-wei
m.s-Lam 3m. s-come to-f. s-part y com 3f.s-one
ka-khinna ta-ba ki-khlm-ya-wan-la
f.s-child only-t hat pl-neg-gm-come-together
Lam came to the party with a girl but they did not come together.
7. *u-lam u-ya-wan-la bad ka-wei ka-khinna a-ka-jikhawai
m.s-Lam 3m. s-gm-come-together com 3f.s-one f.s-child to-f. s-part y
ta-ba ki-khlm-ya-wan-la
only-that pl-neg-gm-come-together
*Lam came to the party with a girl but they did not come together.
We shall demonstrate that it is the occurrence of these two verbal elements (adverb -la-together
and the verbal reciprocal -ya-) that permits the comitative PP to head a relative clause and an EHRC.
We shall now discuss the case of relativization of the comitative PP in sentential relatives.
4.1. Relativization of the comitative PP in Sentential Relatives
Sentence (8) below is an example of a sentential relative in which the comitative PP is relativized. In
this sentence, the head of the matrix clause is ka-khnna f,s-child. The relativized NP along with the
preposition bad-ka-ba with-she-who, that is, with whom is fronted and t
1
(trace
1
) in bold in (8) indicates
the position in which the comitative PP originates.

119
Note that ya-functions as an accusative case marker, a group marker, a hortative marker, a collaborative
effort marker as well as a verbal reciprocal in Khasi.

8. ka-khnna [
S2
bad-ka-ba u-lam u-ya-wan-la t
1 S2
]
f.s-child com-3f.s-adjr 3m. s-Lam 3m.s-gm-come-together
ka-di ka-para (j
120
)-u-kr
3f.s-be f.s-younger sibling gen-m.s-Ker
The girl who Lam came with is Kers sister
The crucial point about (8) is the obligatory occurrence of the two markers the group marker -ya-
and -la together which are not present in the modification of any other PP in Khasi. Recall that these two
markers are not needed in a simple sentence (4) unless the act of the two participants coming together is
focused upon as in (5). We shall argue that it is the incorporation of these two markers in the embedded verb
that enables the comitative PP to be the head of an EHRC in Khasi.
4.2. The Gap relative (Externally-Headed Relative Clause-EHRC)
An EHRC cannot be formed with the comitative PP ka-khnna f.s-child as head as (9) shows.
Note that in (9) the comitative preposition bad with does not overtly occur.
9. *ka-khnna
i
[ba u-ban u-wan
i
a-n
AP
]
f.s-child adjr m.s-Ban 3m.s-come all-village
The girl with whom Ban came to the village
Sentence (9) is an example of a Gap Relative (EHRC) in which neither the relativized NP occurs
nor does the preposition bad with occur. The reason for the non-occurrence of the comitative preposition is:
externally-headed gap relatives universally do not carry the adposition that denotes the case relation of the
head noun phrase with the embedded verb. We have indicated the gap in the embedded clause by
i
and the
gap is coindexed with the subject of matrix clause ka-khnna f. s-child.
Sentence (9) is ungrammatical because the verbal reciprocal with the interpretation of a group
marker and the adverb -la together are not incorporated in the embedded verb. In contrast to (9), (10) is
well-formed because the group marker -ya-and the adverb -la together, both occur with the embedded verb,
just as in sentential relatives (as in sentence (8) above).
Comitative pp as head with a verbal reciprocal ya-and the adverb -la-together
10. ka-khnna
i
[
AP
ba u-ban u-ya-wan-la
i
a-n AP]
f.s-child adjr m.s-Ban 3m.s-gm-come-together all-village
The other predicates that require the group marker -ya-and the adverb -la together in the
formation of a relative clause and a gap relative include: mar run, ya:
y
dki walk, ji-swim among
others. The following examples are illustrative.
11. ka-khnna
i
[
AP
ba- u-ban u-ya-mar-la
i AP
]
f.s-child adjr- m.s-Ban 3m.s-gm-run-together
The girl with whom Ban is running
12. ka-khnna
i
[
AP
ba u-ban u-ya-y
y
dki -la
i AP
]
f.s-child adjr m.s-Ban 3m.s-gm-walk-together
The girl Ban is walking with

120
The genitive marker j is obligatory with personal pronouns, but may optionally be deleted if the noun that
follows it is either a proper noun or a non-specific noun. E.g., ka-kt-j-a my book, ka-kt-(j-)u-bn Bons book,
ka-tnat-(j-)u-d A trees branch.

13. ka-khnna
i
[
AP
ba u-ban u-ya-ji -la
i
]
f.s-child adjr m.s-Ban 3m.s-gm-swim-together
The girl Ban is swimming with
Verbs such as ya-krn converse, ya-th marry
121
, ya- fight, ya-knd meet which are
inherently reciprocal verbs do not require the incorporation of the adverb -la together, and hence, the
occurrence of -la with the embedded verb is prohibited in gap relatives as well as sentential relatives.
14. ka-khnna [ ba u-ban u-ya-krn
i AP
]
f.s-child adjr m.s-Ban 3m.s-Vrec-speak
The girl Ban is talking with
15. ka-khnna [ ba u-ban u-ya-th
i AP
]
f.s-child adjr m.s-Ban 3m.s-Vrec-write
The girl Ban is married to
16. ka-khnna [ ba u-ban u-ya-
i AP
]
f.s-child adjr m.s-Ban 3m.s-Vrec-beat
The girl Ban is fighting with
17. ka-khnna [ ba- u-ban u-ya-knd
i PP
]
f.s-child adjr m.s-Ban 3m.s-Vrec-meet
The girl Ban met with
The other inherently reciprocal verbs that take a comitative PP are -ya-lki play, ya-mi fight (verbally)
We shall argue that it is the incorporation of these two markers ya-group marker and -la
together in the embedded predicate that enables a comitative PP to head an EHRC. Note that a comitative
prepositional phrase which is neither an essential argument nor a subcategorized argument of a predicate
gains the status of an argument that is thematically-linked to the embedded predicate due to the incorporation
of these two markers.
The question that arises now is: What is a thematically-linked argument and what makes an NP
eligible to head a relative clause and a gap relative in Khasi? First, let us enumerate the positions in which a
relative clause and the gap relative occur in Khasi. These positions include the subject, the direct object, the
indirect object, the object of a preposition (ablative, instrumental, locative and comitative) and the object of
the genitive.
For an NP or a PP to be able to occur as an argument in a clause, it should be either a subject or an
object (direct or indirect) or a PP that is thematically-linked to the predicate of the clause.
A question that arises now is: what makes an NP eligible to head an EHRC in Khasi? Can the notion
of subcategorization
122
, for example, play any role in making an argument of the embedded predicate eligible
to head an EHRC, or is there some other factor? We wish to show that subcategorization has no role to play,
but it is the syntactic phenomenon of incorporation that operates on the embedded predicate that make an
argument eligible to head a comitative PP.

121
Note that th means write and ya-th means vrec-write that is marry, which literally means, to write
to each other. This might have started with the tradition of writing/having to write an agreement on paper to solemnize a
relationship.
122
Chomsky (1965) used the term subcategorization. Subcategorization in simple terms means the number of
essential arguments that a verb can take. As Lasnik and Uriagereka with Boeckx (2005: 3) put it: (a) lexical entry must
contain syntactic information about whether, for instance, a verb is transitive, in transitive ,ditransitive, and so on (so-
called subcategorization). See Chomsky (1965) for further details.

Before we proceed further to explicate the syntactic phenomenon involved, we shall first mention
some facts related to the issue. There are two specific sets of verbs that can take a comitative PP in a clause.
Set 1 includes embedded verbs such as wan come, lyt go, mar run etc. which take a comitative PP
with the interpretation of accompaniment. This set of verbs is in no way linked to the embedded verb either
thematically or in terms of subcategorization. Set 2 includes embedded verbs such as ya-krn talk, ya-th
marry, ya- fight which are inherently reciprocal verbs and they are not verbs indicating accompaniment.
In such sentences, the subject of the embedded clause and the head of the comitative PP are in a reciprocal
relationship. The object that occurs with such a set of verbs is a subcategorized argument. Furthermore, in an
EHRC, the argument of the embedded clause that heads an EHRC must be in a specific grammatical relation
with the embedded predicate. The arguments that qualify universally to head an EHRC include subject, direct
object, indirect object, object of the genitive and the oblique object that includes locative, ablative and
instrumental PPs.
A comitative PP can head an EHRC in South Asian languages if and only if either:
(i) postposition Incorporation in the embedded verb takes place in a verb-final language, or
(ii) the verb carries a marker that indicates some group activity together with the occurrence of an
adverb with the interpretation of together, or
(iii) the head of an Internally-Headed Relative Clause (IHRC) carries overtly the comitative PP
(Subbarao 2010 a).
Let us examine if the notion of subcategorization mentioned above is of any relevance in accounting
for the modification of the comitative PP. As mentioned above, the NPs that can head an EHRC are: Subject,
Direct Object, Indirect Object, Oblique Object (Locative, Ablative, Instrumental, etc.). Note that Subject is
an external argument of the embedded verb and hence, it is not a subcategorized argument. Direct and
indirect objects are subcategorized arguments of the embedded verb.
We now look at the Oblique Objects. Locative PP is a subcategorized argument in case of predicates
such as keep, place, live etc. Predicates such as sleep, sit , walk, run etc. that take a locative PP too are
thematically linked to the predicate of the embedded clause though they are not subcategorized arguments of
the predicate.
The instrumental PP and the ablative PP are not subcategorized arguments of the embedded verb,
but they too are thematically linked to the (embedded) verb as the occurrence of the instrumental and ablative
PP is crucially dependent on the nature of the predicate. A predicate such as cut
123
or slice for example
invariably takes an instrumental PP and predicates such as get down, alight, bring for example require a PP
which is source or an entity that potentially moves.
We now consider the comitative PP that denotes accompaniment. The comitative PP is in no way
connected/ related to the embedded predicate either in terms of subcategorization or thematic relation.
Predicates such as come, go, walk, swim can take a comitative PP. In fact, that is the reason why the
comitative PP does not form the head of an EHRC in any South Asian language and some genetically
unrelated languages such as Japanese and Korean unless: (i) the embedded predicate carries a special marker
or markers that indicate accompaniment and some group activity; or (ii) the comitative PP as head in an
Internally-Headed Relative Clause (IHRC) carries an overt comitative postposition; or (iii) the embedded

123
Let us consider (i) from Khasi.
(i) ka-tari
i
[ ba u-ban u-t ya-u-s

i
] ka-l ka-ba-l
f.s-knife adjr m.s-Ban 3m.s-cut acc-m. s-fruit 3f.s-be 3f.s-adjr-blunt
The knife Ban cut the fruit with is blunt.
Note that in (i) there is no special marker that occurs with the embedded verb that shows the relationship
between t cut and tari knife. However, the predicate t cut and the NP tari knife are thematically linked as the
verb cut requires an instrument such as knife for cutting.

predicate in an EHRC carries an incorporated comitative postposition (Subbarao 2010 a). Thus, such special
markers include either an incorporated postposition in the embedded predicate as in Mizo, Hmar or Thadou
(Tibeto-Burman) or the comitative postposition that overtly occurs with the head of a comitative PP in an
Internally-Headed Relative Clause (IHRC) as in Sangtam or Tenyidie, or Konyak (Tibeto-Burman) or the
occurrence of a group marker and an adverb that indicates togetherness in the embedded verb as in Khasi
(Austro-Asiatic) and Manipuri (Tibeto-Burman) (Subbarao 2010 a, c).
Regarding the case of the comitative PP in Khasi, as we have seen, a comitative PP cannot head an
EHRC. The ungrammaticality of (9) above is illustrative. However, when the verbal reciprocal functioning as
a group marker and the adverb indicating together occur with the embedded verb, the sentence is
grammatical as in (10)-(13).
The question that arises is: What makes the head of the comitative PP eligible to head an EHRC in
Khasi? We wish to claim that just as an instrument such as a knife, or saw is thematically linked to verbs such
as cut, slice, saw, the occurrence of the group marker and the adverb indicating together with the embedded
verb makes the object of the comitative PP eligible to get thematically-linked. Thus, the Thematic Eligibility
Condition (TEC) is met and thus, it is this thematic linking that makes the comitative PP qualified to head an
EHRC in Khasi.
The object of the genitive is not an argument of the embedded verb and hence, we ignore it for the
present discussion.
5. Evidence in support of Incorporation in the embedded predicate in a comitative PP in an EHRC
We present below three pieces of evidence in support of our hypothesis concerning thematic linkage.
(i) The first piece of evidence comes from the grammaticalization of the verbal reciprocal -ya-and the
adverb -la-together to form a new predicate. These two elements together constitute a predicate in Khasi
that imparts the meaning of meet / get together/ gather in one place / date someone as in (18).
18. ya-la
vrec -together
meet / get together/gather in one place/date someone
When the nominalizer ji-occurs with the predicate ya-la v rec together, it imparts the
interpretation of meeting.
19. ji- ya- la
nozr- vrec - together
meeting
The fact that the predicate ya-la has the interpretation of get together/gather in one place / date
someone shows that the interpretation of togetherness is imparted by the combination of -ya-and -la, and
that is precisely what helps the comitative PP to head the EHRC in Khasi.
(ii) The second piece of evidence comes from inherently reciprocal predicates such as ya- fight, ya-krn
talk, ya-th marry, ya-knd meet, ya-kdup hug/embrace, ya-d kiss which subcategorize for a
direct/indirect object which may take a comitative postposition. Recall that these verbs do not indicate
accompaniment. In this set of reciprocal predicates, there are a few predicates that also take an accusative
object as well in which case they are examples reflecting group activity and not a reciprocal activity in Khasi.
We found only two predicates so far which take either an accusative or a comitative postposition. These are
ya-th marry and ya-knd meet.
20. ka-ida ka-ya-th ya-/bad- u-ban
f.s-Shida 3f. s-marry acc/com m.s-Ban
Shida married Ban.

21. ka-ida ka-ya-knd
124
ya-/bad- u-ban
f.s-Shida 3f. s-meet acc/com m.s-Ban
Shida met Ban.
With such predicates when a comitative PP heads either a sentential relative clause or an EHRC,
only the verbal reciprocal -ya-which is inherently present with the predicate occurs and the adverb -la-
together cannot occur. Sentences (22)-(25) are illustrative
125
.
22. ka-khnna [ba u-ban u-ya- ] ka-di
f.s-child adjr m.s-Ban 3m. s-vrec-beat 3f. s-be
ka-paralk j-a
f. s-friend gen-1s
The girl Ban is fighting with is my friend
23. ka-khnna [ba u-ban u-ya-krn ] ka-di
f.s-child adjr m.s-Ban 3m. s-vrec-speak 3f. s-be
ka-paralk j-a
f. s-friend gen-1s
The girl Ban is talking with is my friend.
24. ka-khnna [ba u-ban u-ya-th ] ka-di
f.s-child adjr m.s-Ban 3m. s-vrec-write 3f. s-be
ka-paralk j-a
f. s-friend gen-1s
The girl Ban is married to is my friend.
25. ka-khnna [ba- u-ban u-ya-knd ] ka-di
f.s-child adjr m.s-Ban 3m.s-vrec-meet 3f. s-be
ka-paralk j-a
f. s-friend gen-1s
The girl Ban met with is my friend
In case the predicate is inherently reciprocal, the specific occurrence of the adverb -la-together
with the embedded predicate imparts the interpretation that the subject of the embedded clause together with
someone else is performing the action with/for/against the subject of the matrix clause as in sentences (26)-
(28).
26. ka-khnna [ba- u-ban u-ya--la] ka-di
f.s-child adjr- m.s-Ban 3m. s-gm-beat-together 3f. s-be
ka-paralk j-a
f. s-friend gen-1s
The girl who(m) Ban along with someone else is beating (someone) is my friend.

124
The verbal reciprocal ya is grammaticalized and hence, ya-knd is lexicalized as one word.
125
Note that the marker ya-in these sentences functions as a verbal reciprocal marker, and not as a group
marker.

27. ka-khnna [ba u-ban u-ya-krn-la ] ka-di
f.s-child adjr m.s-Ban 3m. s-gm-speak-together 3f. s-be
ka-paralk j-a
f. s-friend gen-1s
The girl who(m) Ban together with someone else is talking to is my friend.
28. ka-khnna [ba u-ban u-ya-th-la ]
f.s-child adjr m.s-Ban 3m. s-gm-write-toget her
ka-di ka-paralk j-a
3f. s-be f. s-friend gen-1s
The girl Ban together with someone else is marrying is my friend.
[That is, Ban and some other person are getting married to the same girl on the same
occasion.
126
]
(iii) The third piece of evidence comes from a set of languages that permit the comitative PP as head
(Let us call it Set 1) as against the other set (Let us call it Set 2) that do not.
As we mentioned above, the comitative PP is not linked either syntactically or thematically to the
embedded predicate. Hence, unless the embedded predicate carries a special marker that establishes a
relationship with the head of the comitative PP, a comitative PP cannot head an EHRC.
Set 1: Dravidian and Indo-Aryan languages belong to Set 1. Dravidian languages do not permit the
embedded predicate to carry a specific marker such as together and hence, none of the Dravidian languages
permit an EHRC with a comitative PP as the head with the interpretation of accompaniment, though an
oblique PP such as locative, ablative and instrumental can form an EHRC in Dravidian (Subbarao ms. &
Subbarao 2010 b). The following examples with the comitative PP as head in Dravidian are illustrative.
EHRC with the comitative PP as head not permitted
Malayalam (Dravidian)
29. *bbu vanna eu
Babu come. adjr letter
The letter with which Babu came. (Asher and Kumari 1997: 61)
Tamil (Dravidian)
30. *kumr vei ye pna um
Kumar out go.pst. adj r Uma
Uma, with whom Kumar went out. (Annamalai 1997: 78)
Telugu (Dravidian)
31. *nnu ve-in-a saraa
I go-pst-adj r Sarala
Sarala with whom I went. (Ramarao 2003: 79)

126
In the existing social set up, it is not possible for two persons to get married to the same girl but that is what
the interpretation of the sentence is.

Kannada (Dravidian)
*huuga (-jote) banda sula 32.
boy (-with) came.pst. adjr Susheela
Susheela, who the boy came with (Sreedhar 1990:58)
Most of the Indo-Aryan languages do not permit the modification of an oblique PP as head. Though
a few of them like Marathi, Konkani, Nepali and Oriya (IA) do permit the modification of oblique objects,
they do not permit the modification of the comitative PP as head (Subbarao, in press). Cambridge:
Cambridge University Press.)
Marathi (IA)
33. *karua ge- le-l-i mul gi
Karuna go- Perf Pple. f girl
The girl with whom Karuna went. (Lalita Dhareshwar p.c.)
In most of the Tibeto-Burman languages too a comitative PP cannot head an EHRC (Subbarao, in
press). In Manipuri (Tibeto-Burman) the comitative PP can head an EHRC provided:
(i) the verbal reciprocal na occurs with the embedded verb, and
(ii) the adverb min together occurs with the embedded verb.
EHRC with the comitative PP as head
Manipuri (Tibeto-Burman)
34. tomba-n lak-min-na-b nupi-du pha-i
Tomba-nom come-toget her-vrec-nozr girl-def good-[-fut]
The girl with whom Tomba came is good. (Subbarao, in press)
35. tomba-n pha-mi n-na-b nupi maca-du magi macanupi- ni
Tomba-nom sit-together-vrec-nozr girl-def his- daughter- is
The girl with whom Tomba sat is his daughter. (N. Pramodini, p.c.)
In Manipuri the reciprocal marker na together with the incorporated adverb min together imparts
the interpretation of doing an act together. The other verbs that require both min and na are verbs such as:
stand, swim, crawl, walk, run, cross the river, jump etc. just as in Khasi.
In Japanese and Korean a comitative PP may head an EHRC if the adverb with the interpretation of
together is incorporated, just as in Khasi and Manipuri.
Japanese
36. [matt - ga *(issyoni) kita] onna- no- ko
Matt- nom together came woman- of- child
The girl Matt came with (Masayoshi Shibatani, p.c.)
In Korean too the occurrence of the adverb hamkkey together with the embedded predicate is
obligatory in an EHRC with a comitative PP as head.

Korean
37. yeongsig- i *(hamkkey)- on yeca
Yeongsig- nom together- come girl
The girl Yeongsig came with (Masayoshi Shibatani, p.c.)
Other intransitive verbs such as swim, walk, run, sit, crawl etc. too permit the comitative PP to head
an EHRC if the adverb together occurs with the embedded predicate in Japanese and Korean (Matt
Shibatani, p.c.). The occurrence of the adverb together in Japanese and Korean clearly demonstrates that a
comitative PP can head an EHRC if and only if the predicate can get thematically related to the embedded
predicate.
In German (Tatiana Oranskia, p.c.) and Russian (Ludmila Khoklova, Boris Zakharyn & Tatiana
Oranskia, p.c.) a comitative PP cannot head an EHRC as these languages do not permit adverb incorporation
just as Dravidian languages cannot.
Thus, both Khasi (Mon-Khmer) and Manipuri (Tibeto-Burman) employ an identical process in the
formation of the Gap Relative with the comitative PP as head, namely, the addition of the verbal
reciprocal/group marker and the adverb denoting together to the embedded verb to enable a comitative PP
to head an EHRC.
Consequently we may reconsider the case of the comitative PP in Khasi. The predicates wan come
in (10), mar run in (11), ya:
y
dki walk in (12), and ji swim in (13) are not thematically linked to the
head of the comitative PP. That is the reason why a sentence such as (9) repeated here as (38) is
ungrammatical in Khasi.
38. *ka-khnna
i
[ba u-ban u-wan
i
a-n
PP
]
f.s-child adjr m.s-Ban 3m.s-come all-village
When the group marker -ya-and the adverb -la-together occur with the embedded predicate, a
predicate such as wan come gets thematically linked to the head of the comitative PP. Hence, a comitative
PP qualifies itself to become the head of an EHRC. If a predicate is inherently reciprocal, only the group
marker which also functions like a verbal reciprocal is required and the occurrence of the adverb -la-
together alone is sufficient to establish a thematic link.
Therefore, it is the thematic link that permits either an instrumental PP or a comitative PP to head an
EHRC in Khasi. Our generalization also has predictive power; a predicate such as ji swim or pr crawl
which does not get thematically linked to the head of the comitative PP obligatorily requires -la together .
Our predication is rightly borne out in (39) and (40) in Khasi.
39. u-khnna [ ba u-ban u-ya-ji-la ] u-di u-paralk j-a
m.s-child adjr m.s-Ban 3m.s-gm-swim-together 3m. s-be m. s-friend gen-1s
The boy Ban is swimming with is my friend.
40. u-khnna [ ba u-ban u-ya-pr -la ] u-di u-paralk j-a
m.s-child adjr m.s-Ban 3m.s-gm-crawl-together 3m. s-be m. s-friend gen-1s
The boy Ban is crawling with is my friend.
Our analysis has general predictive power too. A comitative PP cannot head an EHRC or an IHRC
in any language unless the embedded verb carries a marker that indicates accompaniment in an EHRC or the
internal head in an IHRC carries an overt adposition. Further, if it is a predicate which is [+inherently

reciprocal] in a language, the adverb together may not occur. Manipuri
127
and Khasi provide evidence in
support of this claim.
An inherently reciprocal predicate such as khat fight or luhong marry does not require the
adverb min together to occur with the embedded predicate. Only the verbal reciprocal is required.
Manipuri (Tibeto-Burman)
41. tomba- na khat- na- ba nupi maca- du eigi i manabi- ni
Tomba- nom fight- vrec- nozr girl- def my female friend- is
The girl who Tomba fought with is my friend. (N. Pramodini, p.c.)
The fourth piece of evidence comes from clefts in Khasi. In clefts the clefted noun phrase is fronted
and a new CP node is created to accommodate the clefted noun phrase and the verb di be in Khasi. The
clefted noun phrase carries the preposition that it carries in a simple clause.
An example of a simple clause with a locative PP is given in (42) and the corresponding clefted
sentence is given in (43).
Khasi (Mon-Khmer)
42. u-wan u-b ya-ki -kt ha-lr-ka-mc
m. s-Wan 3m. s-put acc-pl-book loc-surface-f. s-table
Wan puts the books on the table.
43. di ha-lr-ka- mc ba u-wan u-b ya-ki -kt
be loc-surface-f. s-table comp m. s-Wan 3m. s-put acc-pl-book
It is on the table that Wan puts the books.
When a comitative PP is clefted, as expected, the occurrence of the verbal reciprocal/group marker -
ya-and of the adverb -la-together is obligatory as in (44).
44. di bad-u-paralk j-a ba u-ban u-ya-ji-la
be com-m. s-friend gen-1s comp m.s-Ban 3m. s-gm-swi m-together
It is with my friend that Ban is swimming.
However, if the predicate is inherently marked reciprocal, only the verbal reciprocal -ya-occurs and
the adverb -la together does not as in (45).
45. di bad-ka-paralk j -a ba u-ban u-ya-mi
be com-f. s-friend gen-1s comp m. s-Ban 3m. s-vrec-fight (verbally)
It is with my friend that Ban is fighting (verbally).
Thus, the occurrence of the markers -ya-and -la in clefts parallels the one in EHRCs with the
comitative PP as head. This further substantiates our claim that the comitative PP as head requires the verbal
reciprocal -ya-and the adverb -la with predicates that are [-reciprocal] and only the verbal reciprocal -ya-
with predicates that are [+reciprocal].
To account for the facts, we propose the Thematic Eligibility Condition (TEC) which states that:
Each predicate in the embedded clause of the EHRC has to meet Thematic Eligibility Condition
(TEC) for it to be able to accept a comitative PP as head. Such eligibility is the result of language-specific
syntactic processes such as Postposition Incorporation, the incorporation of the verbal reciprocal as a group

127
Thanks to N. Pramodoni for a helpful discussion on this issue in Manipuri.

marker and the adverb together or the incorporation of the adverb together alone in the embedded predicate
or the overt presence of the comitative postposition with the head in an internally-headed relative clause.
(Subbarao, in press)
6. Conclusion
In this paper, we have shown that the comitative PP as head in relative clauses and EHRCs in Khasi
requires the verbal reciprocal -ya-and the adverb -la-together with the embedded predicate, when the
comitative PP is not thematically related to the subject of the embedded predicate. If it is an inherently
reciprocal predicate, then the occurrence of -la-together is prohibited, as it imparts the interpretation that
the subject of the matrix clause and the head of the comitative PP are together performing the action
for/against someone else.
Thus, for a predicate to able to license a comitative PP as head of an EHRC/IHRC, the embedded
predicate must be thematically eligible/ qualified. Such eligibility is achieved by any one or two of the
following processes that effect the embedded predicate:
(i) The presence of an adposition (due to incorporation in the verb or occurrence with the head in an IHRC),
as in Tibeto-Burman languages,
(ii) The addition of an adverb such as together to the predicate as in Japanese and Korean and Manipuri
(Tibeto-Burman).
In Appendix 1 and 2 below we provide data from Khasi which illustrates the various positions that
can be relativized in Khasi.
Appendix I
Relative Clauses in Khasi
Subject Modification
(1) u-brw [u ba la-y ya-ka-pisa ha-u-khinna?] u-di u-paralk j-a
m-human 3ms adjr pst-give acc-f-money dat-m-child 3ms-be m-friend gen-1s
The man who gave the money to the boy is my friend.
Direct and Indirect objects Modification
Direct object
(2) ka-pisa [(ya-)ka ba u-brw u-la-y
f-money acc-3fs adj r m-human 3ms-pst-give
ha-u-khinna?] ka-di ka ba la?-jt
dat-m-child 3ms-be 3fs adjr perf-torn
The money which the man gave to the boy is torn.
Indirect object
(3) u-khinna? [(ha-)u ba u-brw u-la-y ya-ka-pisa]
m-child dat-3ms adjr m-human 3ms-pst-give acc-f-money
u-l u ba jr
3ms-be 3ms adjm tall
The boy who the man gave the money to is tall.

Oblique object Modification
Locative
(4) ka-m
y
d [ha -ka ba a-bo? ya-ka-kt] ka-la?-kd
y
a?
f-table loc-3fs adj r 1s-put acc-f-book 3fs-perf-break
The table on which I put the book is broken.
Ablative
(5) ka-jaka [na-ka ba u-(la)-wan ] ka-l ka-ba ji
f-place abl-3fs adjr 3ms-pst -come 3fs-be 3fs-adjm far
The place which he came from is far.
Instrumental
(6) ka-tari [da-ka ba u-(la)-t ya-u-s?] ka-l ka-ba l
f-knife instl-3fs adjr 3ms-pst-cut acc-m-fruit 3fs-be 3fs-adjr blunt
The knife with which he cut the fruit is blunt.
Comitative
(7) ka-khinna? [bad-ka ba a-ya-wan-la]
f-child com-3fs adj r 1s-vr-come-toget her
ka -di ka-para (j)-u-meban
3fs-be f-younger sibling gen-m-Meban
The girl who came with me is Mebans sister.
Object of the Genitive Modified
(8) u-brw [u ba ya-ka-y j-u la-pinjt da-ka- r y] u-da-ym
m-human 3ms adjr acc-f-house gen-3ms pst-destroy instl-f-storm 3ms-prog-cry
The man whose house was destroyed by the storm is crying. Lit: The man he that his house was
destroyed by the storm is crying.
(9) u-brw [u ba ka-y
m-human 3ms adjr f-house
j-u ka-(la)-sha-pinjt ha-ka- r y] u-da-ym
gen-3ms 3fs-pst-SA-destroy loc-f-storm 3ms-prog-cry
The man whose house got destroyed in the storm is crying. Lit: The man he that his house got
destroyed in the storm is crying.
Object of comparison
(10) u-brw [u ba u-lan u-kham-jr (ban
m-human 3ms adjr m-Lan 3ms-more-tall than
ya-u)] u-di u-paralk j-a
acc-3ms 3ms-be m-friend gen-1s

The man who Lan is taller than is my friend.
(11) u-brw [ya-u-ba u-lan u-kham-jr] u-di u-paralk j-a
m-human acc-3ms-adjr m-Lan 3ms-more-tall 3ms-be m-friend gen-1s
Appendix II
Participles in Khasi
Subject Modification
(1) u-brw [ba y ya-ka-pisa ha-u-khinna?] u-di u-paralk j-a
m-human adjr give acc-f-money dat-m-child 3ms-be m-friend gen-1s
The man who gave the money to the boy is my friend.
Direct and Indirect objects Modification
Direct object
(2) ka-pisa [ ba u-brw u-y ha-u-khinna?]
f-money adj r m-human 3ms-give dat-m-child
ka-di ka ba la?-jt
3ms-be 3fs adjr perf-be torn
The money which the man gave to the boy is torn.
Indirect object
(3) u-khinna? [ba u-brw u-y ya-ka-pisa ]
m-child adjr m-human 3ms-give acc-f-money
u-di u ba jr
3ms-be 3ms adjm tall
The boy who the man gave the money to is tall.
Oblique object Modification
Locative
(4) ka-m
y
d [ba a-bo? ya-ka-kt] ka-la?-kd
y
a?
f-table adj r 1s-put acc-f-book 3fs-perf-break
The table on which I put the book is broken.
Ablative
(5) ka-jaka [ ba u-wan] ka-l ka-ba ji
f-place adjr 3ms-come 3fs-be 3fs-adjm far
The place which he came from is far.

Instrumental
(6) ka-tari [ba u-t ya-u-sO?] ka-l ka-ba loo
f-knife adjr 3ms-cut acc-m-fruit 3fs-be 3fs-adjm blunt
The knife with which he cut the fruit is blunt.
Comitative
(7a) ka-khnna? [ba a-ya-wan-la] ka -di
f-child adj r 1s-vr-come-toget her 3fs-be
ka-para (j)-u-meban
f-younger sibling gen-m-Meban
(7b) ka-khnna? [ba ya-wan-la bad-a]
f-child adj r vr-come-together com-1s
ka -di ka-para (j)-u-meban
3fs-be f-younger sibling gen-m-Meban
Object of the Genitive Modified
(8) * u-brw [ ba ya-ka-y j-u
m-human adjr acc-f-house gen-3ms
la-pinjt da-ka-r y ] u-da-ym
pst-destroy instl-f-storm 3ms-prog-cry
The man whose house was destroyed by the storm is crying. Lit: The man he that his house was
destroyed by the storm is crying.
(9) * u-brw [ba ka-y j-u ka-(la)-sha-pinjt
m-human adjr f-house gen-3ms 3fs-pst-SA-destroy
ha-ka-r y] u-da-ym
loc-f-storm 3ms-prog-cry
The man whose house got destroyed in the storm is crying. Lit: The man he that his house gt
destroyed in the storm is crying.
Object of comparison
(10) * u-brw [ba u-lan u-kham-jr (ban ya-u)]
m-human adjr m-Lan 3ms-more-tall than acc-3ms
u-di u-paralk j-a
3ms-be m-friend gen-1s
(11) * u-brw [ ba u-lan u-kham-jr ] u-di u-paralk j-a
m-human adjr m-Lan 3ms-more-tall 3ms-be m-friend gen-1s

Abbreviations
1 first person;
3 third person;
acc accusative;
adjr adjectivalizer;
all allative;
com comitative;
def definite
DO direct object;
f feminine;
fut future tense;
gen genitive;
gm group marker;
inf infinitival marker;
IO indirect object;
loc locative;
m masculine;
nom nominative;
nozr nominalizer;
OO oblique object;
perf perfective aspect;
pst past tense;
s singular;
SA self-affective
VRec verbal reciprocal;
References
Annamalai, E., 1997. Adjectival Clauses in Tamil. ILCAA, Tokyo University of Foreign Studies, Tokyo,
Japan.
Asher, R. E. & T. C. Kumari, 1997. Malayalam London and New York: Routledge.
Bareh, C., 2007. (A) Descriptive analysis of the Jowai and Rymbai dialects of Khasi, Ph.D. thesis, North
Eastern Hill University, Shillong, India.
Chomsky, N., 1965. Aspects of the Theory of Syntax. Cambridge, MA: MIT Press.
Lasnik, H. and J. Uriagereka with C. Boeckx. 2005. Minimalist syntax. Malden: Blackwell.
Ramarao, C., 2003 Telugu velugulu (Telugu Brightness in Telugu). Hyderabad: Navodaya Book House.
Sridhar, S. N., 1990 Kannada.London: Routledge.
Subbarao, K.V. South Asian Languages: A Syntactic Typology. ms. (in press).
Subbarao, K.V. 2010 a. Comitative PP as Head in Internally-Headed Relatives Clauses in South Asian
Languages. Keynote address presented at International Conference on South Asian Languages
(ICOSAL) at Punjabi University, Patiala, India.
Subbarao, K.V. 2010 b. Dravidian Languages: A Syntactic Typology. Keynote address delivered at the
seminar on Language and Genetic Architecture of Dravidian at Dravidian University, Kuppam,
India.
Subbarao, K.V. 2010 c. Tibeto-Burman: A Syntactic Typology. Keynote address delivered at the seminar on
The Languages of Nagaland at Nagaland University, Kohima, India.
Temsen, G. M., 2006. Embedded Clauses in Khasi. PhD thesis, Delhi University, Delhi.
Temsen, G.M. 2007. What is the Order? Paper presented at the National Seminar on Emerging Linguistic
Scene in North East India. North Eastern Hill University, Shillong, India.

Icaal 4-1

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Icaal 4-1

Uploaded by

Copyright:

Available Formats

Pacific Linguistics E-8

[2DU-PROG-roast.PFV fish] they were roasting the fish

). Since these instances were rare, it is impossible to determine whether they

s na, gd yl, (5)

b to have specks in ones eye: an undergone experience

r, ajoor common tree-shrew

d fruit bat, flying fox

d young child (plural sdy

d, but no singular *sy

l glowing ember: it keeps glowing of its own accord

soc scarlet fever (Diffloth, see in Isara

r, population primitive de lOuest du Cambodge. BEFEO 41: 1-150.

r, population primitive de louest du

-seg mmr-seg sun-

jh (Khmu Rawk) and pllh (Mlabri);

h (Khmu Rawk) and paph

.woman hear FIN IRR.say FOC be FOC manner TOP

in Burmese and rao in Mon.

mtta) and Mon to mean bless, but here

o, which in Mon means bean, pea,

, meaning be up to (ones wish), depend on (someones wish). The literal

.to Lord older.sibling think EXHORTIVE

:/ foul smell, stench

he/she/it is not visible

You might also like